On 43rumors plenty of commenters keep saying that there is a “4 times factor” difference between Full Frame and Micro Four Thirds. For those people a f/2.8 PRO zoom from Olympus actually has an “equivalent” Full Frame aperture of f/5.6. Now, while I agree that this is in a very abstract theory correct… it is actually absurd and wrong if you look at the full aspects of the camera technology.
And without me having to explain you why it is so just read the super well written “”Sensor Crop Factors and Equivalence” article written by from Nasim Mansurov on Photographylife. Particlualry point 8) Total Light explains you why that “virtual equivalence” made by so many bloggers and commenters is de facto nonsense:
8) Total Light
“Equivalence” created another ugly child: total light. This theory, which is brought up by some photographers, says that smaller sensors get less total light than larger sensors just because they are physically smaller. That’s how they explain that small sensors have worse noise performance / overall image quality. That a full-frame sensor looks cleaner at higher ISOs than say Micro Four Thirds, just because its sensor area is four times larger. I don’t know where these theories originate from, but I fond the idea of “Total Light” and its relevance to ISO absurd. Explaining why sensor of one size has a cleaner output when compared to a smaller sensor just because it is physically larger has one major flaw – it is actually not true once you factor in a couple of variables: sensor technology, image processing pipeline and sensor generation. If larger sensors did receive more “total light” than smaller sensors, then every full-frame sensor made to date would beat every APS-C sensor, including the latest and greatest. Consequently, every medium format sensor would beat every full-frame sensor made to date. Is that true? Absolutely not. Just compare the output of the first generation Canon 1DS full-frame camera at ISO 800 to a modern Sony APS-C sensor – have a peek at this review from Luminous Landscape and have fun comparing. Surprised to see APS-C beat full-frame? No, not really. Newer sensor technologies, better image processing pipelines and other factors make modern sensors shine when compared to old ones. Simply put, newer is better when it comes to sensor technology. APS-C has come far along in terms of noise performance, easily beating first generation full-frame sensors in terms of colors, dynamic range and high ISO performance. CMOS is cleaner than old generation CCD that struggled even at ISO 400! Until recently, medium format cameras used to be terrible at high ISOs due to use of CCD sensors (which have other strengths). But if we look at the theory of “total light”, medium format sensors are supposed to be much better than full-frame just because their sensor sizes are bigger. But if we look at high ISO performance and dynamic range, it turns out that it is actually not the case. So these folks now add a few words / disclaimers at the end of their statements like “as long as the sensors are of equal efficiency and generation”. When we talk about aperture or shutter speed, there is no such thing as a new generation aperture and shutter speed, and yet they think they can slap on those words for sensor performance. Don’t be fooled by such statements, as they make no sense. The theory of “total light” is too darn confusing, so it is not worth wasting time on.
Thanks Nasim Mansurov for bringing down that discussion to a very “practical – reasonable” level 🙂
345 Responses
All that talk and no comment about total light being simply determined by the aperture diameter… misses the point.
LOL, funny article indeed! Thanks for that!
I don’t think this article will settle the discussion at all…if anything the FF acolytes will have a field day in the comments about it!! No argument is anywhere definitive in the article about the light gathering equivalence, and, as darrask says, it misses the point in my opinion.
One thing is interesting though: he uses sensor generation as the defining factor. Following this reasoning, which is relevant indeed, I would ask to the FF crusaders: with all this talk about light gathering equivalence, and how you just shift ISO and shutter speed/aperture to compensate for the differences, has your photography been so horrendous in the past when ISO (ASA) wasn’t anywhere near what it can be now? Or did you just take great pictures with the equipment and performance that was available at the time? Right now, M4/3 has IQ that you could only dream about in the past, whatever the sensor size. But pictures were great then as they are now, only now they have better technical quality. That bigger sensors have potential to give better technical quality pictures in not something anybody can argue about, but is it that relevant when you are out taking pictures? As long as the sensor output quality is good enough for the user, all the arguments lose any importance. Sensor size is only one of many factors for the choice of a camera system, and whoever is buying into one has weighed those factors in already. And if the choice is made by a novice, IQ is not going to be so different to be of any importance to a novice’s eyes. Not with the modern sensors.
All I would ask to the FF and light gathering equivalence crusaders and the trolls (but they won’t give a damn…) is to respect our choice of system and sensor size!
Yes, we know, FF has better sensor output generally. So what?
Why are you asking for respect, no need for that. I like my FF-manual-focus setup for mechanics, the shutter slap sound, the DOF. Thats why I still have some stuff from there. My main camera (intended as secondary system previously 🙂 ) is my E-M1 with a couple of primes and I like it very much for the purpose of making photographs, no more photo romance there without the old mechanics. But what I wanted to say, what do you mean with IQ is better from FF? I see no style related to sensor size and cropping and zooming on screen is better with FF indeed but that does not make IQ. I wish the discussion would aim a bit more on the output medium: and that is printing fotobooks for me. Paper contributes more to the IQ than we would think and larger sensors are useful for larger prints. Just that the relevance starts to show from sizes only that might be uninteresting for the most of us. Maybe the discussion should concentrate more on these relevant things about what we want to do with our captured moments and how. Then sensor size is one factor out of a bunch and not even the most important. Thats my prayer…. and maybe the reason, why I find Nasims blog one of the more valuables out there.
I’m asking to respect the choices of gear each of us makes, but I’m not asking you: I don’t need to, you already have respect.
I’m asking to those who criticize indiscriminately any other system that is not their choice and that feel so superior with their (so much better in their own eyes) choice, not taking into consideration that other people have a right to a different opinion to theirs.
I do find Nasim’s article very interesting and relevant, but I fully expect a lot of people to keep on discussing the topic (and not in a civilized way in many cases) forever. I am bored to death by that same concept being reiterated with no end…reading comments used to be fun and interesting, now it is a struggle to find the intelligent and/or interesting opinions in the sea of pointless light gathering and DOF discussions, which have nothing to do with taking good pictures!!
Give me a break
What about showing some respect first by not calling the any-other-sensor-size-than-m43-users “crusaders” and “trolls”?
You sound like a trolling crusader…
Jesus…bad day? I specifically said the “FF and light gathering crusaders” to distinguish between people that keep on carrying forward those arguments like preachers on this site from normal people commenting and discussing general topics! I never generalized against “any-other-sensor-size-than-m43-users” as you put it!
no bad day at all. You started defending yourself without even being attacked, I just wonder why…
I have nothing to defend, I was just trying to prevent another interminable “lecture” going on for ages on subjects that have been worn to death…and didn’t succeed at all, I see…
Then why do people still clai
Hubertus, with all respect, I distinguished between trolls and “crusaders”…I don’t mean for crusaders to be an offensive term, but you are one of those I was thinking about when talking about “light gathering crusaders”: again, don’t take it as an offense, it is just a fact to my eyes. I understand your argument, you are right in principle, but it really makes no difference to me and you in the field: we adjust the exposure triangle based on the situation and the tool’s capabilities and limitations (including sensor size S/N ratio), without having to make particular calculations every single time we use a different tool! You automatically adjust your settings in relation to what you have available to you (camera/sensor, lens, etc…). What I’m trying to say is that, at the end of the day, if you know what you are doing it really makes no difference in the field what sensor size and light gathering capability you have: you already made your considerations before heading out of the door!! Or, even earlier, in the shop!
On the other hand, if you don’t really know what you are doing, it doesn’t really matter until you mastered a bit of the craft already and finally start understanding what this argument is all about!! At that point you will already be able to make more informed decisions on what to buy and what is suitable for you…
The practical point is, for me: this horse has been beaten for so long there is no horse left, so I would try to find other, more interesting subject to discuss and to learn about…you have a point but lots of people don’t get it or simply don’t think it is so important.
I really hope to have kept this on a friendly tone, I don’t want to be offensive with the “crusader” thing.
What do you think?
I see, so ok, no problem!
The thing is, many people don’t seem to have made those considerations before they go out the door or before they buy a camera, and by listening to people like the one cited in this article they will become misinformed regarding the basics of what sensor format difference really means and what its practical results are.
And I ask myself, why aren’t those told they’re beating a dead horse who start the discussion by publicly advocating falsehoods about how to compare different sensor formats, falsehoods which are able to grossly misinform people who don’t already know better? Why address those who object, if they happen to come across those false statements?
I understand whre you come from, but have you noticed any change in attitude and ideas in those who claim, as you say, falsehoods? Given the fact that this subject comes up so often and that mostly the people commenting here are always the same means that probably the information doesn’t come through. And again, I really think that on a practical level it’s not that important, as long asyou use the right settings for your exposure, whatever the equipment.You’re wasting a lot of energy for not a lot of results, IMHO!
I’m not nearly wasting as much energy as the person who wrote the article that was mentioned in this 43rumors post. And I still wonder why he, and you, put so much energy in trying to tell others that they were wasting energy and that something wasn’t important.
For me, to understand things has always been important and will hopefully continue to be, whether it is important “on a practical level” or not. And maybe this is more important on that practical level then you allow. I, for one, very much believe that to understand how cameras based on differently sized sensors compare does help people to understand their own cameras, even if they will never care about other ones. (And personally I’ve met quite some people who will stay what we’d call beginners their whole life, and they know that and are very happy with it, and they’ll probably keep their three or four years old camera with that one single kit lens at least for another five years, and they still sometimes want to know and they ask me how that camera compares with that other camera. And that’s good, isn’t it?)
Totally agree with the admin on the article!
Well, this may be true for ISO performance.
But if we speak of depth of field :
I love 4/3 compact system, but, yes, I have to admit that with my lovely 45mm 1,8 olympus, I take the same image than a FF would with a 90mm F3,6, concerning depth of field !
(but, that is still awsome :p )
It is not true for ISO performance, either, as everyone knows who’s used both small-sensor and large-sensor cameras of a similar technological level.
like a smartphone and a medium format camera? If you want to make claims, you better look for extremes and check them again…
Yes, like smartphone and MF camera indeed, as long as both sensors are at least roughly on a similar technological level, which of course is and has always been a prerequisite for the rule. Thankfully, nowadays there’s the latest Sony 50 MP MF sensor which perfectly extends the rule into the practice of medium format, too, not just the theory. And the closer the existing technology will come to its theoretical limits, which can easily described by area, timr and photon count, the more ubiquitous will it becomen in practice, too – while even smaller sensor sizes will become “good enough” for many practical purposes.
Another unscientific preacher of the earth-is-a-disc religion of photography, repeating the same old factually wrong and logically flawed arguments which have been invalidated thousands of times in the last few years. He doesn’t “know where these theories originate from”? Well, that’s only one of the many simple things he doesn’t know, and the answer is “the most basic principles of geometry and optical physics”. “If larger sensors did receive more ‘total light’ than smaller sensors, then every full-frame sensor made to date would beat every APS-C sensor, including the latest and greatest” – sorry, but that’s so stupid it hurts. No, it wouldn’t. It only would and it indeed does, as long as both use comparable technology, and that’s why the latest medium format sensor, Sony’s 50 MP chip in medium format cameras by Pentax, Phase One, Hasselblad and the like, perfectly follows the light-gathering-determines-low-light-capablilities rule, too…
Please explain, just saying that the other is wrong does not help much. I found this total light theory irritating too: looking closely through a pinhole gives the same view as looking through an open window, because the ‘mass of light’ does not exist. Light is just light under equal circumstances. The total light theory would mean an Elephant should see better than an eagle, but it is exactly the other way around. I am a layman with layman examples. If I am completely wrong, please explain to my why?
On dpreview, there was a comprehensive article on the subject, written by people who know both theory and practice, perhaps that makes it clearer?
There is no such thing as a “total light theory”. There is just the extremely simple fact that when there is light of a defined brightness shining onto a surface, an area four times the size of another area receives four times as much photons as the smaller area, which is four times the amount of light, in the same time.
And of course it’s the amount of photons received by a light-sensitve area that determines how good an image can be produced, beside other factors, of which the efficiency of the device is one. If both are similarly effective, the larger sensor can either use four times the number of same-size pixels, resulting in four times the resolution with the same pixel-level image quality, or use the same number of four times as large pixels, resulting in better per-pixel image quality (dynamic range and less high-ISO noise), or anything in between.
It’s the same reason why a small-sensor SLR viewfinder, showing the same image at the same F-stop as a full-frame SLR viewfinder, is either much smaller or much dimmer, and extreme measures have to be taken to produce a decent finder image like the Olympus E-3 and E-5 had (which, as a matter of fact, were the only Four Thirds DSLRs with a decently large and bright viewfinder image, and that’s one of the reasons that made them so heavy and bulky).
Thanks!
One thing certainly is true. This theory contributes much to the cardiac activity of some bloggers here… 🙂
Well this theory is utterly nonsense!
To get the 4 times larger sensor fed with enough light it need a 4 times larger opening than the sensor a quarter it’s size!
The intensity per pixel is exactly the same and that’s where the measurebaters get it wrong they think that the intenity of the light received per pixel is higher. gain nonsense
What is an “opening”? There is no such thing in optics. There is the (relative) “aperture” of a lens, which defines its brightness, meaning the light intensity of the image on the focal plane. And then there is the “image circle” which defines the maximum area that can be lit with that brightness, which is probably what you mean. But that is a design aspect of a lens for which there is no number printed on it. It has nothing to do with its relative aperture number (F-stop), nor with its focal length.
Everyone’s missing one thing: quantum shot noise. Photons are random in time and position. The smaller your sample, the more likely you see “noise” generated by randomness.
Noise comes in many forms. One form of it will be most prevalent and dominate the others for any implementation you put together. You can’t zero out all noise.
If what you have said was true, then there would be no point in putting your camera on a tripod and using a slow shutter speed with a low ISO to give lower noise photos in low light.
By using a longer shutter speed, you collect more light, thereby showing how the total light idea is correct
Wrong, this theory applies for all factors equal but sensor size. But give me a break, I have to admit, I am not really interested in finding out the rights and wrongs. The only thing I can say is, that I did not take lesser photos with smaller sensors. But that, of corse, is not science.
I couldn’t agree more with you. His “opinion” is full of flaws and actually explained exactly…nothing
Hubertus here is one of those “crusaders” I was talking about…and probably you might be as well, seeing this comment and how you reacted to my previous post.
Oh, well…
aaaahm what? you need to take your pills little boy, it’s just talking about cameras and not the second world war…
Look, let’s leave it at that, shall we? I’m surely not losing my temper, not even close. I just want to keep it friendly and discuss openly.
If I offended you it was not my intention, as I explained above to Hubertus, I don’t mean any offense with the “crusader” definition…
Good day!
no worries 🙂
So,
if you take a Nikon 810 sensor and cut out a 4/3 sensor from it,
it will be noisier, because less total light?
It would lose image quality. A smaller sensor on the same technological level has either less resolution (if you keep the pixel size) or more noise (if you keep the pixel count) or something in between.
Yes, keep the pixel size. Is it total light that gives less resolution or is it the pixel count?
It is all the same. Keeping pixel size the same, less area = less light = less pixel count.
You would only lose resolution. Speaking of which, I would love an approach like the Sony A7s in M4/3: less resolution, more DR and better S/N ratio…12MP with latest technology would be great for me!! That would become my night stroll camera!
Yes, but resolution is an aspect of image quality just as dynamic range and per-pixel noise is one, and in designing a sensor of a given size you can always trade one for another.
You have to take into account, though, that more pixels ameliorate noise compared to less pixels, given a fixed output size for the image. IMHO, the extreme high-ISO performance of the A7S (compared to the A7 and A7R) is more the result of a targeted optimization of all sensor unit components with regard to high-ISO, not so much the pixel size alone.
That’s why some talk about “per-pixel noise” vs. “per-image noise”.
By the way, I still find myself repeatedly astounded by what the E-M1 is able to do at night, thanks to sensor and stabilizer, shooting things I can hardly see with my own eye!
Hubertus, we can engage in a peaceful discussion without calling each other names can’t we? Let’s try to be a bit more civil, I think it will be beneficial for everyone. I respect your opinion and I thank you for your feedback, I really appreciate it. You brought up valid points and you are taking one sentence and attacking it, while that was not the point of the article.
Using “comparable technology” and assuming that things are “equal” has its own flaws. We cannot assume that all camera systems from all manufacturers will be “equal”, even if we look at the same generation sensors. Sony A7R’s SNR is a bit different compared to Nikon D800’s. Their RAW output (11+7 bit lossy vs 14-bit) is different. Canon’s 5D Mark III has a different SNR compared to Nikon D800. Smaller sensors are getting more efficient for their size. So on and so forth. So why try to compare these systems with a flawed assumption of everything being “equal” when it almost never is the case? Yes, if everything is 100% equal, larger sensor will always yield better performance relative to its size. However, what if a company patents and releases new sensor technology that is far more efficient than what we see today? Say an APS-C sized sensor that has better SNR and dynamic range than full-frame? What then?
My recommendation is to drop these assumptions, because it leads people in wrong direction. A Micro 4/3 user that just got started with the system and has never shot full-frame should not know what his 25mm f/1.8 is “equivalent” to. It is not important for them – it only presents useless and irrelevant information that only ends up confusing them. If this topic was so easy to digest, would you have so many comment and questions from people? Why does equivalence always create such heated discussions? Because it is very confusing for most people out there…
But you are a hack. You are misusing complex technological terms and concepts to lull beginners into a false sense of understanding. It is like a model where you throw out too many variables and couldn’t balance the equations and ended up just falsifying the data. You are no different than those that preach water-memory and other snake oils.
Another anonymous comment with a person who has nothing to say except calling other people names. If you disagree with the content of the article, please bring up your points and I would be happy to discuss…
There are so many other factors besides “Equivalence” in the digital world.
“Equivalence” would make sense if still in the “analog emulsion” era.
Nowadays so many other factors make it a true … mess.
Aside from the very common PP most people here do anyway you also
have the specialized options to de-noise (Neat Image a.o.) and re-size (Photozoom a.o.).
Even a 1″->1.33″ sized sensor, combined with proper extremely fast lenses (e.g. LX100), can produce indistinguisable FF-like results, even in low light etc … given proper PP.
I’m not sure about “indistinguishable”…from what point of view? In small prints, probably. Web use, yes, most certainly, if not enlarged to sizes close to original. For You have to consider that the pixel peeping crowd is quite large… ;-P
Despite being a wholehearted MFT proponent, total light is credible and in fact correct to explain why sensors of the same generation, at the same ISO setting, with the same megapixel count and similar pitch will have different high ISO performance. Simply the photodiods of the larger sensor are larger and less affected by the noise. The incident photon flux is higher thus yielding a better signal to noise ratio.
If the pixel pitch is the same surely the photodiodes are the same, not larger?
Nearly, if you assume the photodiodes have zero distance between them. What I meant to write was distance not pitch, you are right.
When technology advances far enough, the image quality will be almost indistinguishable between sensor sizes, i.e. ISO 3200 will look mostly the same to the naked eye no matter if it is FF or 1″ sensor. The relative difference will remain the same of course, but the absolute amount of noise will become so small it stops to matter. After that, the only meaningful difference between formats will be FoV and the DoF difference that comes with it, and even that will probably be software corrected in the near future. All in all, sensor size will matter less and less.
The basics will most likely remain for a long time 34mm and aps-c winning in quality when a shalow DOF is needed and/or low noise ISO… And 43 winning in weight size when you need more DOF. but in the end these are more whining theoretical discussions, happy with my 43 gear and so is my back
While the principle behind what you say is correct, you seem to be missing the fact that Four Thirds is closer to APS-C than APS-C to 35mm, both in actual sensor size and in high-ISO and DOF aspects.
How close FT and APS-C really are is shown by the fact that my E-M1 beats my 16 MP NEX in terms of low noise & high detail at high ISO, if only slightly, when compared by RAW. The gap used to be large when Olympus and Panasonic were still stuck with their 12 MP sensors which were below par in many aspects, but the current 16 MP sensor technology has finally reduced that the gap to the point where differences between two cameras of the same sensor size may be larger than between a specific MFT camera and a specific APS-C camera.
Well, our Em1 sensor is newer, and lacks of the lowpass filter, thus the high detail performance.
Compare it to a RicohGR (older sensor than Em1, but lacking the low pass filter) and you’ll see how sharp the tiny Ricoh is.
No words about iso equivalence :'(
ISO is ISO is ISO .. but makers do tend to measure it differently.
As it was ever so with film. One maker’s 24ASA would be faster than the next one.
Oh, that means ISO100 was a fast film and ISO200 was extra-fast and …
ISO is indeed a defined standard to create evenly bright pictures across camera’s and formats.Every camera sensor has its own sensitivity, and needs to normalise that to a ISO standard.
As tech inproved our pictures didn’t suddenly get brighter.All improvents made sensors have better noise performance. Thats done by multiplying the signal to the desired value to achieve the right brightness. if a sensor gives a stronger signal, the less boosting is needed to enhance it, and thus less noise.
Thats why its not fair to compare FF ISO 6400 Vs M43 ISO 6400 😉
That relates more to how the JPG is processed. The underlying sensor only has one ISO. When you set a higher ISO the signal is boosted.
When the camera processes the raw file it can manipulate the signal to over or under boost it (effectively) to preserve highlights or increase brightness in the shadows in the final image.
The ISO standard relates to the brightness of the final image, so as long as that meets the standard how you go about getting there is a moot point..
That’s why all these arguments about manufacturer’s ‘cheating’ are nonsense and the argument is rapidly becoming irrelevant with modern ISO less sensors.
Just set the shutter speed and aperture you require and adjust the ISO afterwards to get the image brightness you require (if you shoot raw) with no penalty against setting it in the camera.
Damn, a 150mm f2.8 gathers the same amount of light.
The sensor get the same amount of light but when you compare a 16mp 4/3th sensor and a 16mp ff sensor the pixels of the ff are bigger and more light sensitive that makes the ff a better sensor than a smaller sensor.
Yes but that sensitivity is measured in EV and the latest M43 now have 13 stops much the same as the best FF from Canonikon. Don’t sit on laurels.
The only thing he did was trying to explain why others are so totally wrong, with a very flawed argumentation. Then he actually explained absolutely nothing. What was his point? That the sensor noise on a 2014 m43 camera might be as good as on a 2006 FF camera? That’s great, but it explains nothing about total light. He simply doesn’t understand it.
Does he actually want to tell us that the total sunlight falling on a football field will be the same as the one in his little backyard? Or is it the different kind of gras planting technology?
I think the sun is the same, wether it shines on a football field or in a little garden. I think that the quality of grass is important (pixel size), andif there are clouds in the sky (good glass).
yes, you get the same sunburn in the garden as on the football field.
But:
if you want to plant a certain number of palm trees (Megapixels) on both the football field and the little garden, you will find that they will grow much faster and bigger on the football field whereas they spend each other shadow and block the sunlight in the little backyard, grow slower and don’t get the same high quality (pixel noise and dynamic range) and nice yellow (color range) bananas in the same time (shutter speed) as on the (FF) football field… 😉
That’s what I was “saying”. Some People keep writing about things without taking into consideration all the elements.
(Let’s not forget that gear needs a good photographer to take a nice/interesting photo)
hehe yes, but when we’re talking about gear we can’t simply distract and say “it doesn’t matter” because only the right moment and they eye of the photographer matter. Even most professionals wouldn’t agree, a tool is a tool and it’s important to every artist…
Sure, and that’s why the “sensor size war” is silly: as long as the photographer and customers are happy, there’s no real reason to get mad about gear.
I have never had a customer asking me what kind of gear I use.
I always ask customers what do they need.
If they need a 3 mts wide photo I need to borrow an Hassie from a friend, but so far it never happened.
Sure and there are blowhards that think that any of this engineering mumbo jumbo makes any, but the most fleeting difference, when you’re actually out capturing images.
Shoot, chimp, adjust settings and shoot again. 😉
It reminds me of high-end hi-fi audiophiles, that listen to the way their equipment performs, but not to the music.
I once told a condescending hi-fi dealer that even if I spent a million dollars on a stereo system, it’ll NEVER sound like I’m sitting in Carnegie Hall.
If you shoot multiple formats it’s quite handy to know about equivalence to choose the right lenses, settings, etc.
But you would would consider everything before going out, and know your equipment…if you have to top and think about all those calculations before shooting you will come back home with very few pictures!
In doing so you inevitably use equivalence, even if it is just experience telling you so. I for myself want to understand, too.
Experience is part of it, understanding it is fundamental to really make the right choices. I’m not saying it’s pointless to know and understand it, but then you get past it and start concentrating on pictures!
Sure, but you’d be thinking about field of view with a particular lens/camera combo and looking at the exposure (remember the light meter! 😉 ), not whipping out a scientific calculator.
What you wrote there isn´t about total light, but about pixel density
yes, but you have to agree that a larger number of light “rays” or “photons” or whatever light is will hit a large field compared to a small garden, given the brightness of the sun is the same.
So that explains the total light.
Now we have to talk about the pixel density, because that’s what you also need to make equal to get an “equivalent” picture.
The reason why many m43 sensors look very good compared to the FF sensor (same generation) is that the number of pixels is much smaller (18MP vs 36 MP). Of course the the difference in noise won’t be 2 stops any more but only around one stop, you just don’t get as much detail…
As you said in another post, the only thing that matters is the skills of the players 😉
Actually there comes a point where a larger sensor produces more noise.
The larger sensor has more sensitivity but there is a trade-off between
performance and noise floor, plus the four-thirds chaps have been
working on development: there is far more development going on for small
sensors than big ones..
The top current M43 cameras now have little
more noise than the current top-end FF cameras. This is evidenced by
their actual performance and capability, an M43 can take presentable
pictures after a DSLR has packed up and gone home for the night.
The amount of light seen by any camera is determined by its lens. The lens
aperture is a collimator that regulates this amount of light. Both large
and small sensors are at their respective focal plane and so receive
the same amount of light from that lens.
if you use the same FF lens (lets say a vintage rangefinder lens) with the same aperture on a FF and a m43 sensor, the image circle drawn by the lens inside the two cameras will be of the exactly same size and will have exactly the same amount of “total light” brought into the camera.
The thing is, the FF sensor records most of this large image circle while the m43 sensor only records the middle part of it, while the rest is being reflected or absorbed by the plastic around the sensor
So you still think that smaller sensor captures the same amount of light as the bigger sensor, with the same lens and the same aperture setting?
Important is the S/N ratio. Even if read noise could be eliminated, shot noise always remains. So the S/N ratio scales like the square root of the signal. A larger sensor gathers more photons and will always lead to a better S/N ratio. The benefit in read noise will similarly go into larger sensors, too. Additionally, the amount of (wiring area)/(light sensitive area) changes with size.
Why would I need a Football field’s worth of sunlight if I’m just trying to grow a garden in my backyard?
Probably wouldn’t, because plants are mostly a fixed size and don’t vary all that much with the size of the yard they’re in.
But if you were spreading solar panels out on the football field versus your backyard, you might find a difference ;~).
Thom, great to hear from you always. I am your huge fan!
Zeisschen, thank you for your feedback, I really appreciate it. The point of the article is brought up at the very bottom – that all this equivalence is not important for making great images. Yes, I did compare noise on an old camera to a new FF camera as an example, but I also specifically talked about a number of other points that make “total light” flawed. Different sensor technologies, different pipelines, etc. We cannot always assume that sensors will behave according to their size – that’s simply not true today. Nikon’s SNR is different than Canon’s on same sized sensors. Micro 4/3 is quite efficient for its sensor size, etc. Equivalence might work with the assumption that everything is equal, but how realistic is that assumption in the real world?
all being equal is just the one and only assumption you have to make when talking about total light and sensor sizes. coming up with different sensor generations doesn’t make it any better…
Total light is a nonsense concept. What matters is the performance of the individual pixels in terms of sensitivity and the signal to noise ratio. Liken it to film. If you took Fuji Velvia 100 in 35mm size and stuck it in an Olympus OM-1, it would perform exactly the same as the same type of film loaded in a 120 size in a Hasselblad. No such thing as a ‘total light’ advantage for the Hassie. An f3.5 aperture on a Hasselblad would give you exactly the same shooting performance and parameters as an OM-1, given the same film (sensor), just a hell of a lot more pixels/grains allowing you to print to larger sizes.
I’m not 100 % sure, but I believe in this article the “light gathering” section is flawed – but guess what: I don’t even bother to think about it. What I see is:
1) If I compare high ISO noise from same sensor generations the larger sensor wins. This might be as well for dynamic range and colour depth.
2) The larger sensor camera produces less depth of field in the same setup (light, distance to subject, money for lenses)
OK. But why the hell did I then make the decision to M43?
According to 1) and 2) above my answer is “good enough”, many other things I prefer in M43.
PS: sometimes I add grain to my pictures in post.
1) the full frame sensor is approx 3.84 times the area of a four-thirds sensor, not four times.Do not get confused with image circle diameter and crop factor: area = length times breadth.
2) the amout of light getting to any sensor is determined by the lens or more precisely, its aperture. That aperture defined as f.number = focal length divided by iris diameter. If you like, think of a lens as a funnel, the hole at the bottom is the aperture, big hole more light, small hole less light. Simple. It has nothing at all to do with sensor size.
right, that’s why the actual picture quality of the latest Sony smartphone with F2 lens is just the same as the Sony FF sensor using same generation cmos technology with F2 lens!!
Oh, wait…
The best answer to all the arguments about noise, is that you have to look at the number of photons hitting a pixel. Nothing else matters, you just have to compare pixel size for the same generation of sensor. Bigger pixel = better signal:noise ratio.
Beside your argument, is it not understandable, that noise is magnified 4x until we reach the size of a FF sensor? But that’s nothing to do with the total light theory.
I found the entire article interesting and It attempts to explain much more than equivalence. Worth reading though it’s a bit dry in places. We rarely get much in the way of technical explanations from camera makers and almost never from Fanboys. Read the entire article.
Total light makes perfect sense.
What I find messing it up is “equivalent aperture” and then meaning “equivalent f-number”.
The lens speed is the result of focal lenght /aperture diameter. So a 24mm lens with an aperture of 12mm will have the equivalent speed of a 48mm lens with an aperture of 24mm.
…..for an equivalent photo, the same(equivalent) size aperture diameter will result in the same DOF. However, f2 on m43 is not equivalent to f4 on FF, so you have to raise ISO by two stops on the FF to lower shutterspeed, resulting in similar noise and same total light….
the thing also is, a 25mm F2 m43 lens would be around the same price, weight and size as a 50mm F4 FF lens. You can get “equivalent pictures and quality” with a smaller sensor, but you have to invest in the big expensive lenses. It’s not that simple any more since there are now FF mirrorless, not just DSLR…
No, f2 is equivalent to f4 (with twice the focal length, which makes both, as you yourself correctly say, not just equivalent, but equal: 12mm/2 = 24mm/4 = 6mm) on FF, because that’s what produces two images which will look as similar as they can get, and it’s not “you have to raise ISO”, but “you always can and of course always will raise ISO”, if that same image quality in terms of noise and detail resolution is what you want.
12mm/6mm is equivalent (equal in value) as 24mm/12mm. 12mm/6mm is not equivalent (equal in value) as 24mm/6mm. The f-number is the result of the lens physical properties, the aperture diameter is not the result of the resulting speed of the lens physical properties. I know it looks like your setting the f-number on a modern camera, but its really just showing the resulting lens speed of the focal lenght and aperture diameter.
You dont “have to raise ISO”, you can also lower ISO on the smaller format, but the difference needs to be there if you are making an equivalent photo.
Yes, I agree. Regarding aperture: We just need to distinguish between relative aperture, which is what colloquially is called aperture, and the real or absolute aperture, the size of which can be determined by the focal length divided by the relative aperture. (Actually, to be exact, it’s not quite as simple, as the entry and exit pupil diameters of the specific lens are part of the equation, but it’s good enough for showing what matters when comparing different formats.)
That’s us:
:http://newcameranews.com/2014/12/01/europeans-take-crappy-boring-photo-of-rock/
100% off topic, but hilarious anyway.
I don’t find it off topic. At least not 100% 😉
The message is that we “waste” so much time talking about rules and gearhead things that we don’t even think about what we have in front of us.
The picture in the (funny) article might be wrong from every technical point of view, but I bet it’s one of the most important photographs taken this year.
The point is that we should stop babbling about things like “mine is bigger than yours” and similar, and try to focus on what we should do: take pictures, be it for fun, passion, job.
Unless we’re confined in front of a desktop with nothing better to do.. 😉
Good point, there is a good rule from the old days: f8 and be there 😉
Yes. By full intent. Keeps me awake during boring desktop days 🙂
🙂 same here.
Manufacturers who make Cameras / Camcorders for “Consumers” often have better success with Products that do not require a Doctorate to operate the thing. Manufacturers who make anything that is too great a challenge for many people find their Product is not well adopted.
This is similar for people who try to explain these things in a simple manner.
Incorrect Terms are used, and so-called Theories are derived by people who really do not understand the manufacturing of the Products, but know more about “getting the simpleton off their back FAST” and/or making a SALE (thus, it is an insult to you; the buyer and fool).
There is a Company called “Point Grey” that buys Sensors and makes Industrial and Scientific Cameras. They just posted a Video about their efforts to test Sensors from various Manufactures using the EMVA 1288 Standard.
https://www.youtube.com/watch?v=errwq8Vg9yM&list=PLWa6uO3ZUweAZ-VXnnBsDDsBbz32BLlYf&index=33
In the text above there is this sentence:
“This theory, which is brought up by some photographers, says that smaller sensors get less total light than larger sensors just because they are physically smaller.”.
The author then goes to pick that to bits.
It is an oversimplified ‘Salesman Explanation’ to the fool Customer, either it explains everything to your _complete_ satisfaction, or you understand the person is telling you to “make your purchases elsewhere, no intelligent life here”.
Yeah, well I guess they could have said that “Old big Sensors are probably not as good as the newest Sensors, even APS-C. Also, that new big Sensors are likely better than all other Sensors, except possibly for Samsung’s newest APS-C Sensor. Some of last year’s Sensors are better than early large sized Sensors.” .
A Manufacturer _could_ make a NEW very large Sensor with a small “Well” and a low “Saturation Capacity” and that would “get less total light than” many of the newer APS-C Sensors.
Just because the manufacturer _could_ does not mean they usually do, but you can check through the Parts Listings available at Sensor Manufacturer’s Websites and find many types of Sensors of various capabilities and possibly find one to match your needs.
Point Grey explains in their Video, in fairly easy to understand terms, how to apply the EMVA 1288 Standard to compare Sensor Data on a Manufacturer’s Chart to your needs and choose the best Sensor for your Camera (assuming you buy a Camera from a Company that will put whatever Sensor you choose into the Camera you buy).
As you may have guessed by now, the big name “Consumer” Camera Manufacturers do not offer this Service. You just look at the thing and it’s price tag.
You ask how big the Sensor is, how fast it takes Pictures, how many can Pictures can it take — asking like you think it is a “Film Camera”.
They (the Salesman) answers like they think that the number of Images held on a specific sized Memory Card will determine the lifetime that the Shutter is guaranteed for (the correct answer to the unknowledgeable question – and an answer designed to ‘get rid of you’, fool).
So the simple answer IS that “bigger is better” because the Salesman is not going to sell you last year’s Camera in most cases and any “Pro” who fell in love with the expensive Camera when it came out has had plenty of time to do the research while waiting for the price to come down.
It’s a ‘pat’ answer to a fool’s question, but I do not know that Nasim (the Author) got that, and so he came up with his ‘better Theory’; which is no where near as good as these efforts to provide a Standard by which Sensors can be measured and compared, the EMVA 1288 Standard.
I recommend a read of the .PDF mentioned on their Video’s Page: http://www.ptgrey.com/Content/Images/uploaded/Point-Grey-IndustrialVisionDays.pdf .
Aside from the way sensor size affects the field of view of a particular lens… my 40+ year old Asahi Pentax 50mm looks like 100mm on my M4/3 camera… the equivalence thing seems pointless.
From what I’ve seen in the “photography” world, there are asshats and there are artists.
Asshats yap on and on about about sensor size and equivalence…
Artists just go out and shoot photos.
This statement I agree with. My sensor is bigger than yours people are insecure. These are typically full frame people with all the tech but no talent. So they try find a reason to feel professional.
Not trying to be a d..k but I think it is the complete opposite. I rarely see any mention of this on other forums. There are a few knuckleheads that come here and try to work everyone up but in general, this forum has a fixation with this sensor stuff.
It’s not the M4/3 users who are fixated on equivalence, but the small cadre of insecure trolls who feel the need to anonymously bash the system.
Unfortunately, there are those M4/3 users who feel the need to defend the sensor and the system, which does nothing but encourage and feed the trolls.
Interesting that you never see these FF trolls bashing APS C on Canikon sites. Can it be that most FF trolls are Canicon users and most APS C cameras sold are Canikon? Even if APS C is much closer in size to 4/3, than to FF, you never see this total light discussion against APS C.
“Interesting that you never see these FF trolls bashing APS C on Canikon sites.”
Really? You probably have missed all those who explain away deficiencies of the CaNikon’s APS-C that it’s all APS-C fault and one has to upgrade to FF to get the “proper” camera system.
This days you rarely read that – or see CaNikon adverts to the effect – but it is a subtext to a lot of technical discussions on the CaNikon forums.
I think you answered your own question. Most of the topics here aren’t M/43 sensor vs others formats. If M43 users were fixated on it, they’d be trolling the APS-C/FF boards.
The better question is why the (few) traditional DSLR trolls are so insecure that they feel the need to come here.
Like I said, yes, there are a few people that come here and stir up the pot. they are everywhere. I choose to Ignore them. What I am referring to is the general atmosphere in different forums. I rarely go to a Nikon rumor site and see threads ending in debates about larger sensors being better than smaller sensor or equivalence etc. This site seems to thrive on bashing DSLRS users. Kind of silly really. These are all just boxes that do the same thing. I use a DSLR and come here because I appreciate all types of gear. I am researching a smaller camera system.
I rarely see m4/3 users bashing anyone. What I mostly see is that even passing mentions of m4/3 cameras or lenses acts as a very potent bait for all kinds of trolls that bash everything related to m4/3, and usually glorify this or that FF system.
This happens again and again on one of the general photography gear website I frequent. Under every review or news item that mentions Olympus or Panasonic, there are hordes of trolls ferociously proving how m4/3 is just a stupid toy system for idiots, and only FF is for ‘real’ photographers. But if you look at comments under reviews and articles on other systems, you will find zero m4/3 users. There is no DSLR bashing from m4/3 users at all. Interestingly, once you actually start looking who’s spewing all those comments, you notice that those are mostly APS-C users. Which I find quite funny 🙂
As for this site, it’s got it’s share of regular trolls. And admin fuels and encourages those silly discussions by regularly posting some non-sense about full frame OM-D. I guess flame wars are good for traffic.
We both agree on the relative number of trolls under the bridge, but I disagree that this site thrives on bashing DSLR users. As I said, if that were the case it wouldn’t end here, you’d see M43 fans trolling the DSLR sites.
Given the relative newness of M43, it wouldn’t be surprising that that there’d be a lot of comparison to APS-C since that’s where a lot of us came from (and still use). But that’s not the same as bashing DSLR users.
Everyone here needs to take a deep breath. Look at scarlet Johnson in the ad at the bottom of the page and take a deep breath….relax. Scarlet Johnson. Deep breath…lol.
there are a few knuckleheads but u don’t see m43 users going to FF forums or rumours sights to argue stupid points in desperation. A lot of FF users moved to m43 for a more suitable solution especially in terms of weight/size and I guess the fact that some FF users (probably those with all the gear and no talent) feel threatened that their lens collection will become prehistoric makes them angry. Not saying m43 will prevail but its a given that smaller sensors with new tech in future will make big things obsolete and not needed.
So, when you go out and shoot photos, you don’t care about the field of view of your shots? In other words, you don’t care what you shoot? For those who do, the field of view they want to capture is one of the basics, often the most important one, to determine what lens to choose in the first place…
Huh? I don’t know how you get that from what I wrote.
Maybe I was a bit unclear, but if you try reading the ENTIRE sentence, minus the lens reference, you might understand.
Here. I’ll make it easy for you…
“Aside from the way sensor size affects the field of view of a particular lens… the equivalence thing seems pointless.”
So, pick a lens based on field of view. How hard is that to get?
It’s still you who doesn’t get it. “The equivalence thing seems pointless” only when you put “aside” the whole notion of “field of view”. If you compare cameras while caring about what you shoot, i.e. if you compare the same field of view on both cameras, the assumptions made under the “equivalence” term hold true. Of course “equivalence” assumes that what you shoot is more or less equivalent, too, otherwise it would hardly make sense to talk about equivalence at all.
Point #8 refutes an argument that I don’t think anyone is making, and I disagree that total light is a concept that is “too darn confusing [to be] wasting time on”, if understanding equivalence is the goal. I’ve found the recent DPReview article on equivalence to be clearer & more concise, with photo tests that demonstrate each concept.
Well, a bit stupid. The larger sensor is better than smaller one – all else equal. And there are sensors of different sizes on the same technology level. Know what? The larger sensor gives better results (unless you mess it up). There is no need for m43 people to be so thin-skinned about it. Meanwhile the sensor are for most things good enough (or even better) and – compared with classic DSLRs – you can pick-up a very competent, leight and small kit.
Yes, but a bigger lens also required.
Thanks Nasim Mansurov for *getting away* that discussion again from the very “practical – reasonable” level that it just barely had reached before.
Well, I had a MFT camera for 1 1/2 years, talking A LOT of images with it, and I was really happy with the camera and its outcome. But I can assure you that a f/1.8 MFT lens is the equivalent of a f/3.6 lens on 35mm format. Really!
I think that MFT folks should really just get over this. There are still more than enough reasons why you can be proud and happy about your cameras and systems.
Obviously, didn’t take enough photos then. It’s just simple physics. a f1.8 lens is a f1.8 lens as far as exposure is concerned. Hence why your EXPOSURE VALUES ARE EXACTLY THE SAME. Why can’t you people understand this? The depth of field is in fact close to a F3.6… but more accurately a 25mm 1.8 lens on m43 has the same depth of field as a 25mm 1.8 lens on FF. Why? OH MAYBE BECAUSE THE FOCAL LENGTH IS ACTUALLY, MEASURABLY, OBJECTIVELY… THE SAME. If I could afford to send you all back to basic highschool physics, I would. Not because I particularly care about gear heads trying to be photographers, but because it’s embarrassing that this many adults can fail so dramatically at something humans have had figured out for a few centuries.
You are making a lot of nonsense arguments there.
Exposure values appear nearly the same because ISO is a man-made scale calibrated per unit of sensor area.
Exposure values are not the same though because fast M43 lenses have far worse transmission (T-stop) than full frame lenses.
Focal length is a meaningless measure without a relation to an image plane size. A 25mm lens on a M43 sensor is a standard lens, while the same focal length on full frame is a wide angle and needs a completely different lens design.
Get a better physics education.
Exactly. I used my GM1 and 12mm lens as a light meter for my 1950’s medium format Roliflex last month. Worked great.
That’s juuuuust a bit more “sensor” size between those two cameras.
Only if Circle of Confusion is equal.
I’m probably the guy he’s referring to. He’s right that sensor technology does make a difference, and that’s why I always added the disclaimer that he’s referring to. But it doesn’t make *that* much of a difference.
If you were to crop a D810 sensor to the size of an E-M1, the E-M1 would have 1/12th of a stop better noise. The most extreme example I have (compared to the D810) is the a6000: which has 1/6th of a stop better noise when the D810 is cropped to the same size.
Sensors get better over time, but most manufacturers keep up with each other pretty well (Canon aside, until the 7D Mk II). Yes, new sensors are better than old sensors, and eventually new smaller sensor will produce the same quality as an older larger sensors. Nobody ever argued against this; few buyers are considering either a modern MFT camera or an old full-frame camera.
Comparing the top-rated MFT GH4 to the full-frame D810, the crop factor squared formula would predict that the D810 is 4.0 stops better than the GH4. Measuring the sensors, the D810 is only 3.6 stops better. So total light explains 90% of the difference, and sensor technology explains 10% of the difference.
To my eyes, D810 has only 1 stop or less better than GH4 from 100-800 ISO and slightly more than a stop better from ISO 1600-6400. Even DXO measurement Graph confirms this. 3.6 stop difference is too far fetch.
My measurements are based on DxOMark’s ISO score, for the purpose of accuracy and objectivity. I do own and use both cameras, and the real-world results match up well with my calculations and DxOMark’s ISO scores.
And I don’t see the same chart that you see when I look at the link. The chart I see shows about a 3.6 stop difference (though none of the ISOs line up perfectly). http://www.dxomark.com/Cameras/Compare/Side-by-side/Nikon-D810-versus-Panasonic-Lumix-DMC-GH4___963_943
Wow! You have to scale it. If the GH4 says 791 for low light in DxO, 791×2=1582=> 1 stop. Then, 1582×2=3164 wich means 2 stops more. It is not almost 4 stops!
Doh, you’re right. I meant “3.8 times” rather than “3.8 stops”. I’ll correct it. Thanks for catching that.
😉
If you’re curious, in terms of stops, the GH4’s better sensor technology explains 0.087 stops of improvement over the D810 (based on DxOMark measurements).
Look again. At ISO 200 GH4 all most line up with D810 at ISO400. At 400 GH4 match D810 at ISO800. GH4 @ISO800 match D810 @ISO1600.
Switch back to the default “Print” mode, which is what you need to compare total image quality.
Which ever mode is still the same. D810 shows only slightly more than a stop better at all ISO.
That’s definitely not the case in Print mode, but we can use the ISO numbers to be precise. The D810 is rated at 2853, and the GH4 is rated at 791. That’s a 3.61x difference. The ISO crop factor is 3.83x.
Such imbecilic calculation. Do you know that the ISO is in Log scale?
We use the arithmetic ISO scale, not the logarithmic ISO scale. ISO 100 (arithmetic) equals ISO 21° (logarithmic). ISO 200 (arithmetic) equals ISO 24° (logarithmic).
If you want to know the difference in stops, total light gathered predicts the D810 would be 1.94 stops better than the GH4. Using DxO measurements, the D810 is only 1.85 stops better. The GH4’s better sensor technology amounts to 0.087 stops of improvement.
Looking at screen means you print out the D810 picture way larger than the GH4 picture. Doesn’t make sense to me. T. Northrup is right here.
Note that this chart uses the non-default “Screen” method, which compares individual pixels, and isn’t a meaningful way to measure total image quality. If you select the default “Print” method, the chart lines up precisely with my math, since I used DxOMark as my source for measurements.
I find the definition in terms of “total light” much more meaningful, as I pointed out at photograhylife.com. Scientists usually try to equalise as many parameters as possible before comparing. So here it is AOV, DOF and S/N ratio. The dominant noise contribution is shot noise, which is why the “total light” argument works that well. Read noise gets controlled better, but dominates noise usually in certain areas only (usually in the shadows, see http://www.strollswithmydog.com/comparing-sensor-snr/, for example.).
See DXO graph.
Here’s the graph I see on DxOMark, which supports the math in my other comment (which makes sense, because my math uses DxOMark measurements). I’m not sure why this chart is so different from yours.
Oh, just figured out why your chart is different–you chose the “Screen” method rather than the default “Print” method. “Screen” compares each individual pixel at 100%, but “Print” is the default for a good reason–in the real world we only view images that have been scaled to a given size, whether on-screen or in print.
“Screen” and “Print” aren’t great names for the two methods, since images are scaled on-screen. “Per-pixel” and “Total-image” would be more meaningful names.
SNR is not the problem of modern M43 cameras. The problem is DR and color separation at any ISO.
Its the same graph. Look carefully and see comment below.
Try that at the print option where the 36mp FF will move further ahead and then look at real life images believe me as a GH4 and D810 user I truly wish they were that close. The far higher detail level of the D810 is a major advantage at high ISO as it allows you far more flexibility when processing .There is no need to look at abstract graphs the evidence is in the RAW files
This is the “screen” graph.
It (more or less) shows the difference in performance of a pixel, so the “technology”.
In other words: sensor in D810 is better (both Sony’s, D810 is newer => nothing surprising).
Use “Print” graph to compare whole sensor performance (“total light”). You’ll obviously see a ~2EV difference, because it’s a square root of area ratio.
It’s simple physics and nothing can be done.
The three main differences between sensor sizes and the systems around them (lenses, flashes etc) are: IQ, weight/size and price. We all have different needs and different financial means so it’s natural that some people use smaller sensors and others use larger sensors.
The wisdom is to pick the right system for you and not troll against other people that choosed other systems…
http://photobyandrei.wordpress.com/
Equivalence merely states that “All else being equal’ a sensor four times the size will have two stops better noise performance.
Which is correct.
Except that in the real world you don’t have everything else equal. which may be to the smaller sensor’s advantage (or disadvantage).
Fully agreed! The assumption that “all else being equal” does not work in real life. Smaller sensors are getting more efficient per square inch than their larger counterparts. Too many variables and too many problems with assumptions, which is why I recommend that we do not go there…
Get an Olympus OM-D E-M1 and be happy 🙂
This article is a dud. The ‘light gathering’ discussion is endless because people mix up f-stop and T-stop.
Equivalence using a crop factor works just fine field-of-view and depth-of-field.
The same equivalence for light gathering does not work accurately for two reasons:
– Lenses have different transmission (T-stop)
– Not all sensors are equally efficient
To do a light gathering equivalence you would need to measure the T-stop of each lens and the efficiency of each sensor, and do a far more complicated computation than just multiplying by the crop factor.
The advantage of a larger sensor is very similar to the advantage of a large aperture lens: you get more range.
And that was exactly my point Mike! It is indeed an endless debate, especially once you start looking at differences in real life. And I totally agree with your comments on sensor efficiency and differences in light transmission…
Thanks for the link Admin. Won’t solve anything, debating the merit of “our” favorite system is the only reason for commenting anyway!
As always, “horses for courses” is the rule to follow. Understand what your needs are, find the system with the set of compromises that best suit your needs.
Use whatever system you bought into, and stop worrying about what’s next, or being anxious about your buddy having a “better” one.
There are no bad cameras or camera systems on the market anymore, it’s all good. Better than 99% of the photographers anyway, and I’m no 1
percenter!
When I was old enough to buy my own guns, I read dozens of articles on what would suit me. I saved up, and bought three. A shotgun, a .22 rifle and a .243 rifle. 32 years later, all three still do the jobs they were bought for, and any shots I’ve missed have been my own fault.
I like the guys at PL, but with all that being said, the Four Thirds sensor is still Behind APS-C sensors from Sony and Fuji.
And Noisy. very Noisy, even at 600 iso.
The noise is ok for me, I have more problems with the abysmal color separation at anything above ISO 400.
Yeah, I hope the future sensors resolve the color range and tonality. I think that is more of a pressing issue than ISO performance or DR (as I am pretty content with the DR and ISO performance up to 3200).
Came in to post something like this. The “crop” isn’t the problem I have with the sensor…. it’s the noise at nearly every ISO…
MrALLCAPS, and I never said that M43 was better than APS-C 🙂 I simply stated that we should not focus so much on these comparisons.
For a person who just grabbed a M43 or an APS-C camera, why does it matter that he knows how his camera would be “equivalent” to full-frame? It does not help, just like it does not help for a full-frame shooter to know what his setup is equivalent to MF is. We are confusing a lot of beginners with these subjects and my proposition is to drop this and focus on making pictures. I see where equivalence can be important for a full-frame shooter when evaluating smaller systems (mostly in terms of FoV), but it does not help in any way the other way around…
Oh, nothing against you and your site, I just thought I’d bring that up. And, people will always make comparisons, Light, F-stops, etc.. It’s just going to happen. Having owned M43, APS-C and FF, I feel like Fuji and it’s APS-C sensor has the best compromise out there at the moment. I love my Lumix cameras and the P/L 25mm1.4, But the noise, man…
ALLCRAP, the pixel-peeper…. lol…
I will not read it…
We have exhausted the subject.
Implied in these “total light” references is the Latin, “Ceteris Paribus” (all else being equal). IMO, this is creating a “straw man” argument as to why “total light” people are “wrong”. (A “straw man” argument is an informal fallacy based on the misrepresentation of an opponent’s argument. — http://en.wikipedia.org/wiki/Straw_man )
As Tony Northrup mentioned in his comment, it is silly to think that I would compare my old APS-C 6MP Nikon D70 to my M43 Panasonic 16MP GH3. The D70 maxed out at ISO 1600 and was barely usable at ISO 400+ or so (depending on amount of light or “signal to noise”). My GH3 maxes out at ISO 25600 and is usable to ISO 1600 and sometimes up to ISO 3200 (again depending on “signal to noise”). So even though the D70’s APS-C sensor is “one stop larger” it is at least 4 generations older in sensor technology.
Even comparing the GH3 and GH4, there is about a 1/3 stop improvement in sensor (related) technologies (sensor, processing, etc.), and they were only about 12-18 months apart. I would expect the GH5 (or whatever Panasonic calls it) would be about a 1/2-1 stop improvement over the GH3.
The Sony a7S is a perfect example of sensor technology trumping sensor size… for its current generation of competitors. However, pixel pitch (size) also plays a hand in the performance of the Sony a7S. Some of its performance comes from the fact that its pixels each receive about 3 times as much light as the 36MP Nikon D810 and twice as much light as the new 24MP Nikon D750.
Then again, we are back dealing with “ceteris paribus”. (Math alert!)
According to DxOMark (scientifically, we need “hard” numbers), the low light ISO of the cameras are: D810 = 2850, D750 = 2950, and Sony a7S = 3700. If we were to factor in the size of the pixels, the Nikon cameras would have much better overall ISO sensitivity than the Sony. Also factoring in the pixel size and total sensor size (GH3 @ 16MP, a7S at 12MP), the a7S performs nearly 2 stops better than the GH3/4. (12MP/16MP = .75. 3700 ISO/800 ISO sensitivity = 4.625. 4.625 * .75 = 3.47 times more light per pixel.) About what you’d expect, based purely on sensor size (864 mm sq / 225 mm sq = 3.84 times more light or ~ 2 stops).
Anyway, because of this, I’ve quit worrying about the differences between M43 and FF cameras. I know that my GH3 has very close to the same light sensitivity as the 11MP Canon 1Ds and 18MP 7D, but with better dynamic range.
And, IMO, it totally kicks all Canon DSLR’s video up to the 5D MkIII (at ~4 times the GH3’s current price). And arguably the GH3 can produce better video than some of Canon’s cinema line (i.e. the Canon C100) for ~1/6th the price. The GH4 is even better.
Yes, I’m a fan of the Panasonic GH series. However, that doesn’t mean I wouldn’t love to get a hold of the Sony a7S or a GH series M43 at 12MP (4K x 3K for no 4K video crop and ~25% better light sensitivity) and the latest generation of sensor tech! 😉
Please stop with the glue sniffing lol. Total light is a stone cold fact it works both ways the Sony 1″ sensor performs closer to mFT than the size would suggest due mainly to BSI . The GH3 vs GH4 proposal is nonsense while there has been an advance in video and sensor read etc there has been insignificant advances in IQ it has slightly better DR at the expense of slightly poorer high ISO
http://www.dxomark.com/Cameras/Compare/Side-by-side/Panasonic-Lumix-DMC-GH4-versus-Panasonic-Lumix-DMC-GH3___943_842
Cute. I like the “glue sniffing”. Lol.
Another issue is how these sensors respond differently when taking pics or video. People I respect who have used both the GH3 and GH4 claim that they use settings that are about 1/3 stop better in the GH4 than the GH3. On the other hand, DxOMark found that the GH3 had slightly better low light sensitivity than the GH4 for pics. Of course that better video performance could strictly be better processing, but Panasonic’s engineers claim the sensor is a full stop better. Go figure.
There have been sensor patent filings by both Panasonic and Olympus that indicate we may soon see a real world full stop improvement in MFT sensors in the next generation of two.
It’s a great time to be a photographer/videographer!
Had to stop reading after the first three sentences when it became apparent this guy is a complete and total idiot. How did this rant merit a post?
Orson, thank you for your feedback. Perhaps you should have jumped towards the end of the article, where I say that none of this is really important. I gave a different perspective on the matter and it is your right to disagree. Let’s just do it in a more civilized manner without calling each other names 🙂
Obviously you need to mention sensor tech being equal .If you compare a current mFT to a 9yr old FF { and they do lol} This works both ways for example the Sony 1″ sensor is closer to the performance of mFT than the size alone would suggest. I am sorry to say but the article is wishful thinking fanboy nonsense.
And that’s where the problem is “tech being equal”. We are using total light to compare M43 with full-frame and yet we do not want to look at the fact that M43 is quite efficient for its small sensor size. Nikon D800 has a different SNR when compared to Canon 5D Mark III, etc. Why try to compare systems with a formula that is flawed? What if someone patents a new APS-C sensor that kills all full-frame equivalents tomorrow? What happens with the “total light” assumption then?
Either way, none of this stuff really matters. The point of writing the article was to move away from such discussions and focus on making images, not debating over who is right or wrong on equivalence. I gave my perspective and it is OK if you disagree with it…
The detrimental thing is the amount of light being available to take a photograph. Without light, no photo. SNR is mostly determined by shot noise, nowadays, which is sensor-tech specific. Nevertheless you can, under assumption of equivalence (DOF, AOV,same output size…) calculate the number of electrons being generated by the incident light. This number should be the same across formats, for equivalence. Specifically, this number is dependent on aspect ratio and average effective quantum efficiency (EQE, all other things can be put in a constant, which is the same across formats). EQE varies between 17 to 19% for most sensors nowadays in new cameras I took a look at (exception being the A7s, for example, or BSI sensors, which increase efficiency). This tells us, that “total light” is a very good model to determining equivalence and to get an idea what happens. It helps to understand what is going on and to choose cameras and lenses to go with them accordingly. It furthermore helps to determine cameras punching over their weight (due to a new technology like BSI). Very useful!
Ok, I had to post something because things are getting ridiculously stupid around here. Without bashing I simply don’t understand how so many so said professionals don’t understand this.
First of all we should take in account that each lens is designed for a system. A 70-200 Canon f2.8 is designed for a full frame camera where a Panasonic 35-100 f2.8 is designed for MFT. What does that mean more precisely? It means that the light that comes trough the lens is rated at f2.8 stops and then, trough the lens it’s projected ON THE DESIGNED SENSOR SIZE! I can’t believe that people overlook that.
The thing with equivalence is that people say in videos that MFT (and APS-C for that matter) is a lie, and manufacturers lie to us, that f2.8 is actually f5.6 which is utter bs. Depth of Field, Field of View – yes, those come in a direct relationship with sensor size but light transmission is light transmission.
Analogy time – you have a flashlight in a dark room pointing directly at the f2.8 camera lens fitting it. Your Full Frame sensor will take that light and reduce it to the size of the FF sensor, your APS-C sensor if used with the same lens will have the same amount of light comming trough and so on. If you move to MFT with a MFT f2.8 lens then the light will be taken and made to match the MFT sensor size. If you put the FF lens on the MFT the same exposure will be achieved but a lot of light will be lost because the lens was designed to project on a FF sensor.
I’m tired of people giving conspiracy theories. It’s just sad..
btw – I talked only about f-stop not t-stop, sensor tech etc.
Great explanation, even I understood it! 🙂 You should be the one writing the articles on this topic.
Thanks! I have a typo though “Your Full Frame lens will take the light and reduce it to the size of the FF sensor..”
You two must suffer from a similar kind of brain damage then; this is incoherent garbage.
When talking about strictly F-Stop how is this not accurate? If you have something to add please do so
I can’t find the part where it says anything meaningful to be accurate or inaccurate.
Okay, I think I see what you’re getting at.
Consider a 70mm f/2.8 mounted on a 35mm body, and a 70mm f/2.8 mounted on a Four-Thirds body. They could be the same lens or different ones; all that matters is the numbers and that they both have an image circle *at least* as large as the sensor they’re sitting in front of.
Both of them are receiving the same light from the outside world, and both of them have the same amount of light passing the aperture blades. But three-quarters of the light that reaches the 35mm sensor *doesn’t* reach the four-thirds sensor; it falls somewhere else in the camera body and doesn’t go into making an image. This is a mathematical necessity if the focal length is the same.
So the total light hitting the sensor is then one-quarter as large. Either the four-thirds sensor has the same size photosites as the 35mm sensor — a literal crop, with the same degree of noise but one quarter the megapixels — or it has smaller photosites, achieving higher resolution at the cost of more noise.
In either case, the four-thirds sensor will have a smaller angle of view and produce a different photo — this is why in my opinion it’s more natural to compare two systems with the same angle of view rather than two systems with the same focal length.
Yes, I wasn’t trying to prove anything else. I was just referring to f-stop values because some people consider f2.8 on MFT f5.6 regarding light that passes trough and not depth of field/field of view which is a different story.
Light transmission *per square mm* is equal regardless of sensor size, but since different sized sensors have fewer or more square mm, light per square mm doesn’t tell you anything useful about final image quality.
To calculate depth of field and total light gathered (and thus estimate total image noise), you do need to apply the crop factor to the f/stop.
Note that MFT has a different aspect ratio than 35mm full frame, so the total light calculation is just a bit off–MFT is .062 stops better than the 2 stop estimate.
I agree. There are many considerations when talking about the final image quality: lens quality,sensor tech, sensor resolution, processing methods and even on computer RAW processing methods alter the final product. I’m just stunned that some people heard about equivalence and throw everything up like it’s a giant conspiracy.
I was just talking about basic light coming trough lens on sensor in my post. I know that isn’t all there is for a great picture. Most of the success ( 98%) in photography comes from the man wielding it, but that doesn’t mean any of us should use that as an excuse to get sloppy on the tech side of things.
Go home Tony!!!! Canon is cheating you… ha ha ha….
nooo! not the same amount of light, if the f-stop is the same but the focal lenght is shorter, then it is not the same amount of light! it is the same amount of light per area, since the aperture area is equally smaller as the sensor size!
just imagine two simple lenses (just lenses, not camera lenses) with the same shape but one is smaller (which means smaller diameter and smaller focal lenght, but the same f-stop!!!) if you pair these two lenses with sensors that differ equally in size, they will produce the same field of view. but the smaller lens has the smaller area that can gather light and thus will gather less light!
you are forgetting that the f-stop is relative to the focal length.
also you can think about film, if you take a roll of 400iso film, and cut two pieces one with the area of a ff sensor and the other with the area of a mft sensor and you take a photo with both, you will get the same picture with a f/2.8 lens right? except for the resolution because the “pixel size” stays the same but one piece has just less pixels… obviously you can blow up the mft picture more, but then you will also blow up the noise more which clearly shows, that you had less light….
or if you use the same lens, then you will get a different field of view and this time very obviously less light.
Well I get your point but in your example you are assuming that you would use the same lens (from your film example) which would be a FF lens and yes, you loose a lot of light on the surface it is projected on the sensor of a MFT. The intensity should be the same and that was what I was aiming for in my post.
The best example is a speed booster which acts like a loupe more or less.. it focuses that extra light from APS-C and FF lenses ~ it makes the light spread a bit smaller, thus giving the extra 1 stop of light.
I’m not claiming that Field of View and Depth of Field are the same on a MFT and a FullFrame/ APS-C camera because they are not. That “famous” f2.8 MFT = f5.6 FullFrame refers only to depth of field on calculated focal lenghts for each system [eg MFT DoF 25mm f2.8 = FF DoF 50mm f5.6].
But the speed booster INCREASES the f-stop by decreasing the focal length but leaving the aperture (absolute opening) unchanged. ok that is simplified, but that is the principle. so the speed booster doesn’t make a 25/2 out of a 50/2 it makes a 25/1.4 out of a 50/2! the f-stop is always the same regardles of sensor format. if you take a 25/2 wideangle lens for ff and put it in front of a mft camera it is excatcly the same as a 25/2 for mft, only that the 25/2 for ff is a wideangle lens with a larger image circle. (you should inform you about large format lenses, there you have 90/5.6mm lenses that are normal lenses for 6×9 or 90/5.6 for 8×10″ which are extreme wideangle lenses. (of course you can put the latter lens in front of a 6×9 film but you are wasting a lot of image circle and a lot of money because the 90/5.6 extreme wideangle is much more expensive than the 90/5.6 normal lens.
got it?
I sincerely don’t know where you’re beating at (I probably don’t understand the way that you are explaining).. the speed booster gives 1 stop of extra light and affects FOV by 0.71x and it also affects the DoF.
I found this article to be complete and total rubbish. CCD
sensors are actually MUCH better than CMOS sensors of the same generation for
noise. Astronomers and people who do scientific studies always use CCD sensors
over CMOS sensors because the CCD is much better for that application.
Pixel size is really what matters. The bigger the pixel the
better it is for noise suppression up to a certain point. Total sensor size is
only relevant for depth of field calculations and pixel density calculations.
Neither of those affect noise performance.
Agreed, article such Mr. Masurov’s should not be quoted here, because it
lowers the standard of this site. He should maybe take some classes in
semiconductor physics, quantum theory, solve some of the differential
equations, and get more understanding that a larger surface of a pixel
indeed ensures a better signal/noise ratio.
As it comes to the
“depth of field equivalence”, who is to say that a shallow depth of
field is an universal advantage? This depends on the application, the
kind of photography. As much I enjoy the amazing low light performance
of my FF Canon, when I photograph our girls in low light conditions, in
most cases one of the faces is outside of the depth of field, because
the camera opens up the lens to F4 or F2.8 and that’s that. In fact, I
enjoy the larger depth of field given me in DX or m43 sensors. I always
did, actually, ever since I encountered the smaller sensors. For fast
changing action I found it easier to catch a “keeper” than with a full
frame. But, “when the sun goes down”, the full frame beasts come to
life, and they outshine everybody else… If Mr. Mansurov cannot see
that, he should venture to his optometrist.
I agree with your first paragraph. However, the full frame
camera has a Depth of Field “Control” advantage. I can almost always select a
shallower depth of field if I desire that with a Full frame camera. I can’t
always select a shallow depth of field with an M4/3s camera. The ability to
control the depth of field with full frame is its advantage.
In most cases the Full frame camera will have a larger pixel
area so theoretically it should also have better noise performance even when
you stop down slightly for broader depth of field.
If we look at how sensors are compared today (same print size), there is practically no advantage to larger pixel size. Put a 36 MP D800/D810 image and compare it down-sampled to a D4s with much larger pixels – both will have similar SNR. At pixel level, however, the situation is completely different and that’s where the pixel size matters.
Please see my references to CCD – I actually pointed out that they are inferior at high ISOs specifically and they have other strengths, such as the ones you pointed out. Nowhere did I say that CCD is worse than CMOS in other regards.
when we talk about equivalency we also have to use a FF sensor and a M43 sensor with the same resolution to compare those pictures noise without scaling and stuff.
Then simple physics makes the pixels 3,8 times bigger on the FF camera, that’s why the picture quality is better.
A 100MP FF sensor might be as noisy as 25MP M43 sensor, same picture quality.
No matter the size of its sensor, when I go out and shoot with my E-M1 I have fun, always 🙂
Then that is all that really matters. If you find the experience rewarding and the images you make fulfilling, that’s far more important than any technical aspects. 🙂
I agree, that’s all that matters…
I bet if you found a crop 17 sensor inside your E-M1 after picking it up from cheapcams.com, you would not always have fun 🙂
“If” smaller sensors receive less total light? What is this idiot on about? Is he completely incapable of the most simple math? Probably is, there’s no evidence that he even tried.
A smaller sensor at the same angle of view and same F-number receives less total light, no ifs about it. Less total light reduces the *maximum* possible image quality from a pure information-theoretic standpoint, no ifs about it. A sensor may fall short of the ideal maximum quality (actually all of them do), but assuming equal quantum efficiency, equal performance of ADCs, equal processing algorithms, etc., the total number of photons hitting the sensor is *the* limiting factor for achievable quality.
The way to increase total light is to increase F-number; a 20MPix 35mm sensor with a 50mm f/2.8 in front of it and a 20MPix 35mm sensor with a 25mm f/1.4 in front of it are nearly equivalent systems — and the 25mm f/1.4 will be about the same size as the 50mm f/2.8. The 35mm sensor will be less diffraction-limited though.
If DOF and shutter speed i critical for the final image, all sensors, no matter what size, that have prodused that image will have gathered the same “total light”.
Cheers!
of course, but where do I get an zoom lens for my mft with the same DOF as an f/2.8 lens on ff?
You dont, and there is the FF advantage. Its only in those shallower DOF pictures the FF will gather more total light. If you dont want that shallow DOF, FF will not give you any total light advantage.
that’s exactly where you got fooled by the author of this article 😉 Total light matters, because it’s “just there”
Agree, but I think you could have said it more politely. Your last section has the 35mm sensor twice, by the way, instead of a mft sensor.
Fixed.
Why should the 35mm sensor be less diffraction limited (at least before hyperfocal)?
Andrew, thank you for your feedback, I appreciate it. When I brought up “total light”, I did not say that smaller sensors are better than larger ones. I simply stated that there are many pitfalls in assuming that smaller is always worse in proportion to its size. Unless you are comparing a D800 with a D7000, which has exactly the same performance at pixel level, comparing different systems, different sensor technologies, different image processing pipelines has a number of potential problems. We cannot assume that everything is “100% equal”, because that’s almost never the case in reality. People are trying to compare Micro 43 vs full-frame or APS-C vs other sensors, sometimes of different generations. If we go by “total ligt” as a reference, then Nikon full-frame = Canon full-frame just because both have the same surface area. However, if you look at SNR and dynamic range, the two are quite different, aren’t they?
Either way, the point of the article was not to create yet another debate. We can talk about SNR, ADC and other data all day long, but it will lead us nowhere, because everyone is right and everyone is wrong depending on what side you are looking at…
OMG, you still didn’t get it? Total light is total light, it’s the same for the cameras with the same sensor size and the same lens on it. The difference in picture quality will then be only dependent on sensor technology.
If you use a M43 sensor camera with F2.8 lens and a FF sensor camera with a F2.8 lens (let’s say the lenses have the same measured transmission), then the FF sensor will get around 3,8 times more total light from the same light source on it’s 3,8 times bigger area sensor and thus giving you around a 2 stops advantage in picture quality.
That’s so simple and not confusing at all, your article just made it confusing because you ignore the basic principles and simply say that “they don’t matter”.
What is the CROP FACTOR of the pathetically-small 35mm Full-Frame camera verses the 1:1 Full-Frame Medium-Format-Digitals camera’s robust 53.7 x 40.4 mm sensor ?? Even the 1:1.3 Crop MFD sensor (44x33mm) is huge compared to the Dinky 35mm FF sensor !!
BTW is it OK to say “F*ck Full Frame ??”
no, it’s retarded. Time for bed now..
oh my god, please stop.
to make any statement about this topic we have to simplify. Let’s try please correct me if I’m wrong.
The sensor:
A sensor has the following attributes:
-LST: Light sensitive area (STA (sensor total area) minus the space between the light sensitive area)
-PS: Pixel size (or sensel size)
-ST Sensor technology (very much simplified, that is the part that is responsible for per sensel nosie or signal/noise)
so if we assume (do we know? I don’t know. If anyone knows please step forward), that we can make a mft sensor and a ff sensor with the same LST/STA ratio, same ST and PS, then the amount of light that can be gathered by the ff sensor is exactly that much bigger compared to the mft sensor as the STA is bigger compared to the mft sensor (roughly 4 times).
so far so good. so here the “total light theory” makes sense if we look at two sensors that are exactly the same except for the area.
if we want to compare two sensors that have different PS (pixel size), then we need to know if the PS has any effect on the other factors (LST or ST). If we don’t know that we find out or we shut up ok?
Ok Lenses and depth of field:
Simplyfing:
f-stop is proportional to the t-stop
depth of field only depends on f-stop and focal length
focal lengths written on the lens are true and exact
to further simplify: a lens is just one lens with a focal length and diameter (these two together produce the f-stop (and in our world the t-stop)
so to take the same picture with the mft and the ff camera we need two different lenses with different focal lengths the easiest way is to just shrink the ff lens for the mft camera. so the focal length will change and the diameter will change. and thus the aperture will stay the same! But obviously the mft lens is smaller and will thus gather less light (if you don’t believe this, just get two loupes which are different in size (the greater the difference the easier to see) and try to light a fire. or look at a candle that shines light on a surface area if you increase the surface area, the area (but so that the complete area remains at the same distance to the candle) will still be equally brightly lit but of course the total amount of light energy is greater (the lens is the surface are))
but since the mft lens only has to illuminate a smaller sensor area, the light per area ratio is the same! but since the area is smaller the total amount of light is smaller.
you see the mistake many make is that they forget that the f-stop is a relative size it depends on the sensor size. if we would, in our example, keep the lens diameter the same, then of course the mft lens would have a smaller f-stop which means more light per area.
Oh no… more comments from individuals that couldn’t get a Ph. D. with a thesis on this topic. Why can’t we just all stick to taking photos?
Please do stick to taking photos, instead of joining in just to piss on people who want to have a discussion.
You would be less obnoxious if you took photos.
You would be less obnoxious if you didn’t make unwarranted assumptions about people you’ve never met.
All you have to do to prove your point is to take the camera with the sensors proven to do what you say and take a photo. Post it up with a comparison. Talk is cheap. All that typing about how you know something and calling someone an “idiot” on a previous post is very hypocritical of you in regards to assumptions.
On another note, your comeback was genius.
what are you doing here? taking pictures?
I’m pretty sure discussing sensors is easier than taking photos for you.
You’re probably quite right in guessing that most of those who comment on the technical aspects of photography discussed here could never get a PhD on this topic, but on the other hand, if only people should be allowed to comment on things they’d be able to get a PhD on, most forums would be empty. Then again, that could make things better, overall, one might think 😉
Well said and I agree.
Why aren’t you in the kitchen?
Because I don’t cook. Same as how I don’t ask why you don’t take photos… because you…
Zzzzzz….. just go away…. pls go…………
It’s sad to see that willful stupidity is alive and well. That’s my two cents. I’m going back to making photographs now.
Err… this is weird. We seldom see anyone writing such a long article first to admit and elaborate his lack of understanding on the subject and then went on to argue that people should stop talking about it because they are supposed to be just as confused. But frankly I think the argument is valid, seeing the lot of us still try and debate over the truth and myth.
The problem has always been borderline misleading marketing versus usefulness of equivalence in its incomplete form.
Yang, if you have any particular comments on the article, please bring them up and discuss, once you give it a read. There is a reason why I bring up those topics – to educate first, then to explain that those concepts are irrelevant for most people out there. One cannot just state that something is wrong without first explaining why.
Thanks for the reply. I don’t really see the point, though. There is nothing unclear or needs debating about in this equivalence business from a technical point of view. The problem, in my opinion, includes:
1) Camera maker: “We make this lens that has a bright F/1.8 aperture and is equivalent to 150mm on 35mm format”,
2) Photographer-wannabe: “My sensor/aperture/xxx is bigger than yours”,
3) People who keep denying the validity and usefulness of equivalence based on muddled understandings, and
4) People like me joining these discussions again and again despite better judgement.
My GX-7 could be marked with equivalent values only: instead of 17mm F/1.8 and ISO 200, they could be marked 34mm F/3.6 and ISO 800 equivalent. The comparison of image quality could also be based on equivalent values – the GX7 produces cleaner images than, say, A7, at equivalent ISO settings.
I thought I posted a reply but it just disappeared. Weird.
Has anyone mentioned magnification? Take two sensors, exactly the same technology, exactly the same pixel density. Make one FF, the other APS-C. Put two lenses chosen to match the field of view of equal quality on the cameras and take exactly the same picture (same ISO, same aperture, same shutter speed). The noise captured will be identical.
Make 2 10×8 prints from them and the FF will look better as it has had to be magnified less to get to the print size. The more the image is magnified, the worse it’ll look. Simple.
That’s the factor that most people who don’t understand equivalence seem to miss. It’s the key factor that produces the final results to compare, yet it’s often forgotten.
Exactly.
It is “all about” magnification in the final print. If we do 46×30 print, it will have narrower depth of field than 20×18 print, because we magnify the image more.
When we magnify the final print, we as well enlarge the noise and it becomes more visible by its patterns. But if we have larger pixels per detail, while smaller pixels per detail, then densier pixel has less noise (if otherwise same electronics and generation of technology). But at some point it turns around, the enlarged noise pattern isn’t obvious as it is so large, while the smaller noise pattern is easily spotted.
At some point we can as well lower the required resolution when doing large prints. It is about DPI at that point. No matter how huge resolution we have, we need to recalculate the image to lower resolution to get it printed well. Like if we do a road side boards that are size 7×3 meters, then we do not need 150 DPI or even 75 DPI or even 30 DPI. We can very well use just 2-5 DPI, because the drivers are typically very far away (from tens of meters to even hundreds of meters). So to make a 7×3 meter board with 5 DPI we only need resolution of 1377 x 590 pixels. And how much is that? It is just 0.8Mpix. Yes, 812 430 pixels.
Now how about if we make a typical laptop wallpaper? The most common resolution in typical laptops is 1366×768 and that is ONLY a 1 049 088 pixels, a barely over ONE megapixel.
A Full HD is just little over 2 megapixels. A 4k is little over 8 megapixels. And even with 5k display the screen has barely over 14 megapixels!
If we take a current generation m4/3 camera that has 16Mpix sensor, we could make prints that has DPI of 25 (1 pixel per millimeter) and with that density we could make a print over 4×3 meters (one meter is 1000 millimeters). And that size photo on someones wall is so huge that it isn’t being watched from 1-2 meter distance, and if so it would look awesome.
Now use a 36Mpix sensor and it is just 50% more sharper. Not twice or four times. To get image that has two times more details, we would need to use 16Mpix * 4 = 64Mpix. At that point, each line that 16Mpix captures with one line of pixels, would be captured with two lines of pixels.
Many seems not to even get that the ruling factor for depth of field is the specific print size, viewed from specific distance. And when the image needs to be enlarged more to get to that size, depth of field (acceptable sharpness) gets narrower as the details gets enlarged and we start seeing what is blurry and what is sharp.
If we make print smaller, depth of field gets deeper.
Why can manufacturers sell a bad objectives with plastic lenses? Because most camera users does not do 20×30″ prints but does much smaller ones. Low sharpness doesn’t bother when prints are small or viewed from further distance.
So we welcome the pixel peepers, enlarging the pixels larger and larger and search the differences. Use a FullHD 24″ (22,57×18,84″) display (2Mpix) and enlarge a 16Mpix photo for 100% size (pixel : pixel ratio is 1) and suddenly the photo isn’t 24″ but it is 2,4 x 22,57″ (57,32cm) and 3,2 x 18.84″ (47,84cm) and so it is like 153cm x 137,5cm photo that is observed from 30-40cm distance.
And who does look 1.5m x 1.3m size photo from 30-40cm distance?
How about those who analyze the image files with 200-400% enlargement? They don’t even get that they are looking from close distance a multiple meter photo!!
Would they go 30-40cm of a 7×3 meter road billboard and analyze its noise or resolution patterns? NO! but they do that at home front of their computer!
People seems not to get that the enlargement has so huge impact to everything. It is magnification via distance, cropping (focal length), viewing from computer screen and then finally viewing from print. It is as simple as moving a candle closer the subject or further and get different density of light on subject. Moving camera doesn’t change the subject exposure as the magnification (distance) takes care it naturally. Why we can take photo of person from 1 meter distance or from 100 meter distance with same exposure value and subject is exactly same way exposed.
Andy, this has been specifically addressed in the article when I talk about print size and the advantages of high resolution sensors. Your quote “noise captured will be identical” is only true for pixel-level performance (100% zoom). Sensors today are not assessed on their pixel-level performance, because people realize that different resolutions yield different size prints. In your reference, a good example would be the D800 vs D7000, which have exactly the same pixel-level performance. But once the D800 is down-sampled to 16 MP to match the resolution of the D7000, it will be obvious that the FF sensor will show better SNR.
You are misunderstanding how this works. The noise per pixel will be identical, but when you actually print it at the same size the image from the APS-C sensor will be more visible due to having to magnify it more to get to your 18×12 (or whatever) print. It doesn’t matter how many pixels are on the sensor, the key bit of information is that you are taking an image formed on a sensor that is sized 36mm x 24mm or on one sized 23.6mm x 15.6mm and magnifying them to a size of (say) 18″ x 12″. Look at this another way, taking an APS-C sensor with 16Mpixels and an APS-C sensor with 32Mpixels, but implemented in the same silicon technology. Take an image with both sensors and print them both to 18 x 12. They will look identical – why – because they have been magnified the same amount – it’s not about the pixels, but about the size you render to.
Different sized sensors only yield different sized prints when you print 1:1. I’ve yet to see anyone print an image that is exactly 20.48″ x 16.3733″ (assuming a printer capable of 300dpi) from a D800. The other thing you have to remember is you don’t see the pixel level structure anyway – you can print at 150 or even 75 ppi and people will not see the pixel structure unless they are viewing it far too closely to see the picture as a whole.
There is also far too much reliance on 100% zoom to either explain or explain away issues. 100% zoom is *only* relevant on monitors and then pixel pitch on monitors distorts that further. I have a 24″ and 27″ monitor here – both of equal quality, but the 27″ has a much finer pixel pitch (by about 30% from memory) than the 24″ monitor. Putting *the same* image on both yields different perceived quality – simply because I am magnifying the image differently.
Its simple……APSC and FF produce much better picture quality and less noise at every iso level. But there just to big for most of us m43 users with very small hands and weak body. Its a compromis.
I use earplugs, problem solved
zzzzzz….
Guys, the point of the article, as it states at the very end, is that all this does not matter. Many of us are spending their valuable time posting their take on the matter and calling each other names. Everyone is right, everyone is wrong. Equivalence might make sense for a person switching from one system to another, but for someone who just started with a Micro Four Thirds or APS-C system, what does it matter? Why should a beginner know what their 25mm f/1.8 is “equivalent” to at the end of the day? All systems have their own pros and cons and what works for me might not work for someone else.
All these discussions are creating too much negativity in our community. There is enough of that already out there my friends, so let’s just close this discussion and focus on making great images!
Sorry Nasim,
There is a right and wrong here. There are reasons why these things happen and why equivalence and noise are important for some. Generally there are several things which interrelate and are difficult to explain clearly and concisely – especially if you don’t fully understand what is going on. Always best left alone if you don’t fully understand IMHO 😉
Andy, without a doubt, they are important for some (probably very few). But at the end of the day, such topics are proving to be more harmful to most photographers than useful – look at how many negative comments are being generated everywhere whenever equivalence is mentioned…
Such topics are harmful when discussed by people who don’t understand them and then dismiss them as unimportant….
Andy, please be specific in your references and we can have a discussion. Otherwise it is pointless and counter-productive.
@nasim
“(probably very few)”
Well, I’d say it’s important for everyone who invested in the more expensive FF cameras in order to get a better picture quality. Do you think people chose them to look “pro” or to impress their neighbours?
I hear it so many times here, people saying FF users are so stupid on wasting money for those “huge impractical” FF cameras (available in same size as Em-1 and smaller than GH4 today) just because they have a small Pen*s and need to compensate for it and being blind for not seeing m43 quailty is already “good enough”.
If I hear this, i know who’s the one who is insecure… Equivalency matter, so does total light. You just have to understand it before you dismiss it as unimportant…
Those cameras are impractical for the vast majority of people who aren’t actually professional photographers. I liken it to someone that chooses to buy a gaming laptop yet only uses the microsoft office suite and an internet browser. I have no idea how old you are or where you live, but I live in the burbs in the US and see countless soccer moms/dads lugging around their DSLRs when they would be much better served by a smaller format… and likely even just a high end P&S like the RX100, LX100 or FZ1000.
Sure, the FF body is the same size as some m43’s bodies, but the bodies are more expensive, the lenses are more expensive and the entire system size is much larger overall.
Zeisschen, just an FYI, I mostly shoot full-frame and only have one mirrorless camera, the Fuji X-T1. I am not here to debate that larger is worse. I think the way that this site referenced my article is what created some of the misunderstandings. Please read the article and assess it as a whole, rather than focusing on one particular paragraph taken out of context. If you disagree with anything in my article, please bring it up and I will be open to discussions.
SAYS IT DOESN’T MATTER
WRITES LONG, EXHAUSTIVE ARTICLE ABOUT IT
Hugh, I felt it was better to explain what I meant rather than just say “it does not matter”. Plus, I felt like putting some of the hard to understand concepts in much simpler language might be beneficial to our readers…
“Why should a beginner know what their 25mm f/1.8 is “equivalent” to at the end of the day?”
Because the Oly/Panny software is going to correct the distortion to make it look like a 50mm equiv
this is no software magic going on, you simply just use the center of the lens that it would be for a bigger format, that’s why there is no distortion
You really miss the point of… everything.
The idea behind “equivalence” is not about showing that sensor size X is better than Y. It’s about being able to compare optical systems. So being against it (and you seem to be) is like being against rulers or scales.
I repeat: OPTICAL. In this idea only 2 things are important: the optics and the area you measure (e.g. the CMOS sensor). How the sensor works is not a factor.
The main problem is: you decided to write an article on a very technical topic. However, you’ve made it without EVEN ONE mathematical formula (and proper illustrations).
Just how was this meant to work?
Maybe you wanted to make it easy to read for non-technical photographers, but you simply went too far. There exists a minimal level of formalism which is essential to preserve correctness of a (more or less) technical publication.
And if you are not prepared to write a proper technical text, you really shouldn’t try at all. You’re a good photographer and you should concentrate on that.
Piotr, the article was not written for geeks. It was written for our readers, who are photographers, not engineers. Maybe I went a bit too far, but it was hard to explain things even simpler than that without touching other topics.
If you disagree with what was said, please point out the specifics and I will gladly be open to further discussions and potential changes where I am wrong…
So only geeks deserve a precise information? 🙂
You wrote this text to spread knowledge or to offer entertainment?
From a technical point of view – everything is so informal and bland, it’s almost impossible to guess if you know what you’re talking about.
Example:
You’ve stated that smaller sensors don’t have to be worse, because the technology matters. And this is essentially true, but so what? Sensor technology is highly scalable in the world we’re talking about (10-100 mm). It will always be possible to make a larger (so: better) sensor. Hence, size really matters.
In a more everyday approach, you write some rather confusing ideas.
Example:
“Why should a beginner know what their 25mm f/1.8 is “equivalent” to at the end of the day?”
Seriously? 😮
You want to do portraits and you want a proper lens for that. You open an album, a book or a web gallery and you notice some desirable photos. You’d like to do something similar.
Thanks to the idea of equivalence, you can easily calculate what lens you need (both in terms of perspective and DoF) FOR YOUR SENSOR.
I mean: we could just dump the idea of numbers and call lenses: “landscape”, “portrait”, “other portrait”, “tele”, “longer tele” and so on.
Would this change the resulting images? No.
Would this make life easier? No!
Piotr, I wrote a reply but the comment was “held for moderation”…
If, as you claim, “Everyone is right, everyone is wrong,” why did you write your diatribe, I mean “article,” in the first place, and why do you now defend your rather indefensible position?
Fulcrum Lever, the article was written for our readers in mind, who are mostly not technical. It was oversimplified for their understanding, rather than giving another engineering piece with formulas and SNR graphs. You can find plenty of that on the Internet.
If you disagree with anything in the article, why don’t you provide your arguments on specific quotes? Bring it up and give your opinion on the matter and let’s have a discussion.
Piotr, I am not sure why my previous long comment was not approved, but I will give it another shot and this time make it shorter.
Yes, size does matter and I have specifically pointed it out in the article, if you were to give it a read. However, it is not always safe to assume that it does. Why doesn’t Sony A7s have exactly the same SNR performance as Canon 5D Mark III and Nikon D810, although their sensors are the same in physical size? Why until recently MF cameras struggled with high ISO performance when compared to not only full-frame, but even when compared to APS-C? They are drastically larger, and yet ISO 1600 on a MF back looked like ISO 6400 on my D800. Based on “total light”, isn’t MF supposed to be drastically better than full-frame? And why doesn’t M43 look exactly two stops worse than full-frame? There are many more questions like that, which make it more complex than a simple “all things being equal” assumption when it comes to measuring performance across different systems and manufacturers. Yes, if everything is indeed equal, total light makes sense. But is it relevant today when we are using this flawed logic to compare different cameras?
Lastly, your example on one looking at a photo and then trying to “calculate” what a similar photo could be reproduced as in a different system sounds a bit…unreal to say it nicely? When was the last time you had someone look at a photo and say “hmm, that shot looks like it was taken with a 50mm f/1.4 on full-frame and therefore I will need a 25mm f/0.7 lens to reproduce that shot? I am a photographer and I know many others. I have never seen such behavior, ever. If I look at a photo, I don’t care what it was shot with and I will never try to do a calculation as you have referred in your example. A M43 shooter that is not looking forward to moving up or down in sensor size does not need to know what their camera or lens is equivalent to…
“all this does not matter” to whom? “focus on making great images!” again miss the point. Actually what people should understand is if your going to talk about photography go ahead don’t bring up gear. If your talking about gear, don’t back track and say what does it matter, it is just a tool for photography. Some people care about gear because there are essentially two hobbies going on. Gear collecting (you might think it is stupid, me too) and photography. Keep them separate. You can’t really tell the stamp collector not to collect specific stamps, and not to care about each stamp since they just tools for mailing letters right?
If sensor size doesn’t matter, have fun with the iPhone and it’s f/2.0 lens.
Seems like that’s the solution for all these guys, constantly complaining about “big, heavy cameras”, how bad it is to have shallow DOF because “no one wants to see only a sliver of something in focus”, how “it’s the photographer and not the camera” (even inside a dark reception hall where flash is inappropriate), etc. Delusional bunch here.
Funny, I don’t remember seeing any of those quotes.
Love the look of those EM-10s tho.
Sorry, but that line of reasoning is absurd. I’m a m4/3 guy all the way, but total light and its relevance to noise in the final picture is central and undeniable. Bringing in different sensor tech is complicating the matter for no good reason when it’s understood and a given that the theory applies “all else being equal”. We know all else is NOT equal a lot of the time–we’re not stupid. But that doesn’t make the theory any less applicable. It gives you a frame of reference and it can be used to show whether a certain sensor is over-/under-performing compared to where it should be based on current tech.
Then he admits it’s all too confusing so don’t waste time on it. Completely pointless and doesn’t further the discussion at all. Ignorance is never an excuse.
But as always the more interesting question is, “Does it matter?” For the vast majority of shooters out there the answer is an emphatic “no” even if they don’t realize or want to admit it. Sensor size is an easy way to convince one’s self that a camera is superior. It’s much more difficult to quantify factors such as ergonomics, ease of use, performance, etc., and those matter a lot more than 1.5x or 2x crop difference in sensor size.
+ 1 for the absurd reasoning, the article was totally flawed and ignorant
but don’t believe that M43 system is the only one that has nice ergonomics and ease of use. You get this in every mount and sensor size today, be it smaller or bigger…
MentatYP, I wish admin did not post one part of the article, which makes it sound like I am saying that smaller is better – it is clearly taken out of context. That’s not what I said. I questioned “total light” and stated that it is flawed and gave my arguments. “All else being equal” is too much of an assumption and it is almost never the case. With the exception of very few examples such as D800 and D7000 that share the same pixel size and sensor technology, assuming that sensors will always perform relative to their sizes has its pitfalls.
To give a specific example, if a company patents amazing sensor technology that nobody else can use for a few years and they choose one smaller format like APS-C. And say that technology is amazing to the level where it can easily beat all modern full-frame sensors at SNR, how is “total light” going to hold then? Sure, “in theory”, we could say that “all things being equal”, the same sensor made bigger would gather more light and be even more amazing. But then we would not be doing that – we would be comparing the new sensor technology with existing offers on the market. We would be comparing the output of that sensor to Sony, Nikon and Canon cameras.
Why am I bringing this up? Because I have a friend that works at a sensor development lab. And he was the one that brought up this particular scenario, saying that the sensor they are working on is far more amazing than anything you see today on the market. The biggest issue is cost of production at the moment, but if they can figure out the logistics, we could potentially see disruptive sensor technology soon. Are they going to sell it to all manufacturers or only to one? Who knows…still, that by itself would create issues with these assumptions. And on top of all this, we know that things can look drastically different if we compare different sensor technologies, or look at dynamic range performance. Looking at DxO data, we can see that dynamic range is not proportional to sensor size – some small APS-C sensors look better than same generation full-frame sensors. We can also see that until MF went CMOS, CCD sensors were quite bad at high ISOs. There are many more examples of such problems with total light equivalence…
Every theory is an approximation. One needs to find out how much deviation is tolerable. If the deviation is just a few percent, it is very useful (as are most models used in engineering. They are never exact, but approximate reality well enough). Here, as I pointed out below, the deviating part in this theory centres around average effective quantum efficiency (AEQE, electrons generated due to light hitting the sensor are linearly proportional to this if you derive the number based on equivalence). The deviation within modern sensors is not too large, as I pointed out the values vary between 17-19%. So for the majority of modern sensors the total light thing is a very good approximation and helps further to determine deviations and quantify gains by these sensors providing a sometime disruptive technology. The A7s, for example, can be identified to provide more than you would expect by just using the “crude” total light argument by sensor size alone. The new BSI sensor will give a small performance increase, too. As a scientist, I always look at the assumptions and approximations done when developing a theory. So for me it is an excellent reference for the average of modern sensors out there. But knowing about the limitations, I can easily take new results into perspective and interpret accordingly, as the theory behind it is clear to me. Here I am with you: total light is very useful, but needs to be used with care. As soon as building blocks change dramatically (like AEQE), you need to take this into account. Just using it like a religious mantra doesn’t help.
HF, excellent points! You clearly understand what “total light” means and it is acceptable to you that it is “good enough” for the sensors we see today. For you, saying that ISO 800 on FF is equivalent to ISO 200 on M43 might be acceptable.
I am a bit different in that regard. I look at a larger system and realize that it should be better, because the sensor size is obviously larger. Just like LF is larger than MF or MF is larger than FF. However, I do not want to start calculating SNR or ISO numbers based on assumptions and deviations. I reject the notion that we should be forcing manufacturers to label their products with equivalent numbers, because those numbers do not make sense and are irrelevant for someone who just picked up a camera. I do not want to get involved in thinking about how much specifically one system is worse than another. It is enough for me to understand that images will be noisier with smaller systems, that I will have to get further away from my subjects due to changes in FoV. That’s good enough for me, and for most people out there. Photography Life is not a site for engineers and scientists – it is a site for photographers. The article was written with that in mind 🙂
Perfectly acceptable for me!
None of this will matter when I enact my plan to destroy all small sensors! All hail FULL FRAME!
That’s a rubbish article… it’s backwards for a start. While ‘equivalence’ is a dodgy measure of performance in any context except DoF (for the reasons mentioned above) ‘total light’ is a physical principle that you can’t argue with… total light is the part that’s undeniably correct. Sure it doesn’t matter so much in the overall picture of camera performance… but to waffle on about how it’s “absurd”, “not true”, or “confusing” doesn’t help at all, it just sounds stupid.
+1 my thoughts exactly
Thank you for your feedback. If total light is a physical principle, how do you explain differences in SNR between same size sensors? Just compare the Sony A7s with the Canon 5D Mark III and Nikon D810. Why do they all perform differently, although the sensor size is the same? If total light is undeniably correct, M43 would always be exactly 2 stops worse than full-frame. But we know for a fact that it is not the case – see the same referenced DxO data and the numbers are all there.
Total light only works when you are looking at something like D800 and D7000, which have the same pixel sizes and sensor generation. It does not work for comparing sensors between different manufacturers, M43 to full-frame, or M43 to APS-C. Sensors are different, ADC is different, even RAW files and how information is stored in those RAW files can be different. So what’s the point of total light equivalence, if things are rarely ever equal in reality?
Read noise is no longer an important issue in new sensors (heck D810 is close to be ISO-less). but the thing that doesnt change with technology development is Shot noise, the more you shrink pixel area, the more this kind of noise you will have.
So a fat man at the beach should tan faster, because he has more skin area. 😉 right
I should also add that this situation is really sad not only because of the Mansurov’s article itself.
“Finally a really great article about “Sensor Crop Factors and Equivalence””
In other words: this text states that smaller sensors aren’t worse, so it is great. It must be.
Seriously, similar things don’t happen on Canon / Nikon Rumors. 😮
Piotr, please find concrete references to where it was stated that small sensors are not worse – I never said that. I said that equivalence is flawed in the way that it is looked at today. Please re-read the article and if you have any disagreements, bring them up for a discussion.
“If larger sensors did receive more “total light” than smaller sensors, then every full-frame sensor made to date would beat every APS-C sensor, including the latest and greatest. Consequently, every medium format sensor would beat every full-frame sensor made to date.”
> this logic is totally flawed. That’s why your conclusion also is that it’s no the case.
“So these folks now add a few words / disclaimers at the end of their statements like “as long as the sensors are of equal efficiency and generation”.
> of course they do, because sensor technology on those medium format cameras was not that far in those previous generations. It is now with the latest Sony 50MP cmos sensor.
“When we talk about aperture or shutter speed, there is no such thing as a new generation aperture and shutter speed, and yet they think they can slap on those words for sensor performance. Don’t be fooled by such statements, as they make no sense. The theory of “total light” is too darn confusing, so it is not worth wasting time on.”
>But wasn’t your point that newer sensor technology can compensate. Why do you now say it’s a foolish statement?
Your whole article was very confusing, you should re-read this part and try to think again what you are actually trying to say. You didn’t proof anything with your words that “total light” is unimportant or even non-existing…
You said that the logic is flawed, and yet you yourself admitted that Medium format was not as good at high ISO until we got to the new CMOS sensors. So until those new sensors were released, the “total light” argument was false for MF?
Again, always assuming that all things are equal is bad. It works in theory and when comparing sensors of the same generation and pixel pitch (such as D800 vs D7000), but for all other cases, it is too flawed.
Lastly, if total light was a fact, we would be seeing identical performance relative to sensor size. Is that the case today? No, it is not! Micro Four Thirds is not exactly two stops worse than full-frame – take a look at the same DxO SNR data and you will see that. Don’t even go that far, compare SNR between Sony A7S, Canon 5D Mk III and Nikon D810 – is SNR identical? No, it is not. So why are we still saying that the three are “equivalent” in SNR just because the sensor surface size is the same? Clearly, there are many other variables involved here, such as sensor technology, pixel size, etc.
You are confusing “total light” (based on photons per mm) with “sensitivity” (the ability of the sensor to detect those photons).
GlueFactoryBJJ, wait a second, any time “total light” is brought up discussions revolve around ISO and SNR equivalence between systems (see DPReview’s or Tony’s material) and when I bring it up I am now confusing photos per mm with sensitivity? No one would be stupid enough to argue that larger sensors receive more light due to their larger surface area – that’s a given, it is physics. We are talking about how larger sensors perform relative to smaller sensors in SNR and that’s where there are flaws. Please re-read what I said.
I read this particular thread three times and you consistently confuse the two terms. Perhaps the problem is that you and those of us that are taking issue with what you are saying is one of terminology. Terminology is very important when discussing a technical subject, even for non technical users.
A misunderstanding of what “total light” discussants are arguing is not a basis for an authoritative “disapproval” of their argument. WE understand our point.
For example, Tony Northrup has clearly demonstrated, with real world pics, that in relatively similar generations of sensors, the noise at a given ISO is overwhelmingly determined by sensor size. DPReview comparisons show the same once the ISO levels are sufficiently high.
By this, do I and others who use the “total light argument” mean that sensor tech isn’t important? Obviously not. Barring a disruptive tech change, within, say, two generations, the evolutionary changes are much less important than sensor size (total light) in determining expected noise at a given ISO level.
Hence my earlier comment regarding your “straw man” argument. You are erroneously arguing a point none of the informed “total light” discussants are making. Your misunderstanding of our point is not a valid position from which to argue.
What? This is like holding the test subject as the control and have everything else vary.
ulter bullshit. please go back to basic of fstop and dof, tstop and light.
Supposedly technical articles should be based on fact and science and not on wishful thinking and pandering. Thumbs down.
Tony Northrup and his nonsense religion here?? He he…
The DSLR guy that made an article telling that Olympus an Panasonic were cheating, and only afterward noticed that Canon and Nikon were cheating too?? Ha ha ha… very funny…
How does this apply to my cat photos???
go back to school, nerd
don’t cyber bully me!!!
your cat will be equivalent to a tiger when your’e using a FF lens on your M43 camera 😉
So stay away from the total light because also tigers can see you in the dark…
he doesn’t have an m43 camera and even know how to use one. the type of guy that can only take pictures with clunky and expensive gear, but keep returning to check what’s going on and to determine when to switch… and to troll, of course… btw a very funny clown, i can’t stop laughing… :-I
Equivalence seems to be a hot topic since MILCs are currently in vogue. The article is ok but kind of just shrugs off the debate by stating that in the grand scheme of photography it actually doesn’t matter what the physics are. To technical photographers, the numbers matter since they need to use them to create images. To creative photographers, they may not need the numbers and can generate images by being at the right place at the right time. Making equivalence to 135 format doesn’t really make a huge amount of sense since 35mm film was never really the dominant format in historical context but most times its used as the yard stick due to digital formats since their inception. While the conclusion to the article is correct in saying that people should use whatever suits them the best, it doesn’t really give strengths or weaknesses of formats and specifically the different models of cameras. I do however agree that the “total light” nonsense is pretty flawed in terms of physics because with photographic equipment the transmission of light is a formula that is calculated by the t stop of the lens, the signal to noise ratio of the sensor, and a combination of settings on the camera that control exposure such as shutter speed, the aperture, and ISO sensitivity. Add in the differences in arrays used to gather light and methodology of transmitting the information from the sensor to processor, and the entire concept of “total light” falls to “my hole is bigger than yours”. Equivalence is just a means to an end and that’s for comparing characteristics between different formats and systems. Everything else is subjective chit chat.
Alexander, if things were so easy to grasp and made total sense, we would not be seeing endless debates on forums and over 2K comments at DPReview on this topic. Equivalence makes sense for those looking at multiple different systems, but for most photographers using a single system, it does not matter. I personally do not see anything wrong with looking at FoV differences, but trying to calculate equivalent aperture or ISO, or introducing other concepts such as total light confuses a lot of people.
That is correct, I’m agreeing with you. It’s not that the concept of comparison is difficult to grasp but the misinformation spread by what may seem to be credible sources that confuse the subject. When we are comparing different systems to get DOF calculations, t stops of light, or angle of view/field of view, we can quantify some of the characteristics of lens + system combinations so that we can get a feel for how a lens + system would look if we were to use it in comparison with another lens + system that we already know. Once people start introducing new nomenclature for what should be standardized nomenclature, we start getting zealots on forums arguing about what system is better based on subjective reasoning. Basically things that sound great like “Zoom with your feet”, “always shoot at ISO 100 because it’s the best”, “larger sensors of course get more total light”, “the best lenses are fast lenses so you should always shoot wide open”. You put out a great article to state some ideas but I feel that there will still be lots of forum/internet debate that may not always be productive. Some people (may not even be photographers) will always have partial information or misinformation and try to use it to form an argument. I think there’s a quote but can’t remember who said it “never try to pick a fight with an idiot because they will always drag you down to their level and they will win with their experience”
Alexander, I agree with everything you’ve said, specifically regarding people who misunderstand things and use all kinds of assumptions in their advice to others. You have some wise advice for me, which I will surely take. Would rather focus on writing something more useful (working on reviewing the Nikkor 35mm f/1.8G FX this week), rather than debate with people who won’t use their real names and cannot bring up arguments aside from accusations and name calling.
I actually had two very authoritative people (for me) read the article before it was published. Both of them are engineers (one works for a sensor lab) and they fully agreed on total light being a bogus assumption. Iliah Borg, who I deeply respect, has stated that “Equivalence is most often just the bragging rights” and irrelevant in the way it is presented today, after he read the article. I am not introducing anything new in my article, just questioning all the unnecessary clutter we are presented with.
I actually do think the concept isn’t that hard but part of the big problem I have seen in forums in understandings is simply that people can’t simply accept that their smaller sensor camera actually does not have the same DR, ISO performance or color sensitivity of the other bigger sensor they are comparing against. And that, hey- that’s fine. It can be more than good enough for a lot of things.
If people didn’t have that ‘heart attack’ of inferiority, they would be able to see the concept much easier, at least to compare formats as the poster you replied said.
Reist3d, the concept isn’t hard for those who understand it fully – you are one of the few. And who understand that assumptions on “total light” are not 100% factual, but are “good enough”, considering what we have today.
I think people get too defensive about their choices. I don’t understand either the guys that say that a larger format is overrated and impractical, or the full-format guys that say M43 or APS-C is crap. All have their pros and cons. There is a reason why I shoot full-frame for weddings, landscapes or wildlife. There is a reason why I took my Fuji X-T1 with a single lightweight lens to a long trip instead of my D810 + big / heavy lenses… People need to accept that there will always be a compromise when going from one system to another, just like there always has been a compromise when going from LF to MF or from MF to 35mm… But why get tangled up with all this equivalence in different areas? It might be relevant to those that get it, but most photographers out there really struggle with these concepts and unfortunately, they get in the way too much. I get so many inquiries about this topic…people even have a hard time understanding basic FoV equivalence, let alone other nomenclature. I wrote the article specifically to address those needs and hopefully point our readers to the article saying “there is all the stuff you need to know, but you also should know that most of it is not important. Go take great pictures and stop worrying about how your pictures would look if you were to switch to a different system”…
I am sorry but something here in this cited article doesn’t make much sense. Sure, a more recent smaller sensor has more advanced tech than an older one and can beat an older one. But I seldom see this as a mantra for people who prefer a bigger sensor for better IQ.
At the same level of technology the bigger sensor will beat the smaller one. This is pretty easy to see. As for Nasim- you mentioning “not everything is always equal” – sure, for particular models, at different times, but given how much dominance (other than Canon for the most part) Sony has in sensor manufacture and market, now more than ever a lot of things are indeed equal in in those variables.
Also the utility of knowing equivalent in focal length and aperture to 35mm / FF formats goes well beyond this- because there is so much literature in photography historically with 35mm film, it’s simply convenient to be able to know what lens / exposure parameter your current setup is equivalent to- so you get a good idea from other textbooks in photography talking in those terms.
Also there are some people who mix system sensor sizes so it’s also good to know.
I think there are some good things to consider from the article, but I don’t think the particular point cited by admin is quite right.
Disclosure: I am currently shooting with the CX format for my street work and APS-C format for paid work. I do not have any FF/FX camera.
Of course bigger is better, all things being equal. I specifically pointed that out at the end of the article and never claimed otherwise. The points I brought up in the article in regards to “total light”, is that the assumption is too flawed if we factor in different variables. “All things being equal” is not a safe assumption, because sensors are different, pixel sizes are different, pipeline is different, even RAW formats can be different. So why are we talking about theoretical equivalence, when in real life things are way more different? Canon’s SNR and DR are different than Nikon’s, even with the same sensor generation. Sony A7R’s RAW files are only 11+7 bit. So on and so forth. Too many issues with such assumptions…
Nasim, you are digging like a blind man in things you barely understand. Let it go.
If you have something to say, bring it up and discuss openly. You said absolutely nothing with your accusation.
“All things being equal” is a *somewhat* safe assumption in terms of direction because as I pointed out the majority of sensors out there are all Sony’s. Canon is the oddball here which clearly has being behind in sensor tech. Nikon uses Sony sensors. Pentax uses Sony sensors. Panasonic sensors are at best just behind Sony’s by a hair. Sony’s A7R’s RAW files are compressed but that doesn’t make them “11 bit” per se- the compression is relative to where the data matters creating more often than not a better recreation of the original 14-bit DAC data.
So yes, the variables you mention matter but in most cases at the same year/time of use, not enough to be an offset really in terms of the sensor RAW data *in general*.
Raist3d, but wouldn’t you agree that basing a theory that is now being applied mathematically by some people to derive equivalence on a sensor technology that could potentially change any time in the future is flawed? What if Sony does not manufacture most sensors 5 years from now? Aptina is making plenty of sensors today. Toshiba makes sensors. Nikon makes their own sensors (in addition to using Sony’s). So does Canon. Who knows what happens next. The assumption could be somewhat safe today, but why should we make comparisons across systems on something like this?
I just personally think that it is enough for someone to understand that there are compromises with smaller systems, just like there are compromises when going from MF to FF. Equivalence in FoV is also OK for those that shoot with multiple systems, because it allows one to understand what lens to use…
“Raist3d, but wouldn’t you agree that basing a theory that is now being applied mathematically by some people to derive equivalence on a sensor technology that could potentially change any time in the future is flawed?”
Not really in the context of looking at a sensor size comparison. You can say “all things being equal” and in the theory that’s true, and in the reality it’s effectively pretty close to true because of what I mentioned.
Canon makes their own sensors, and that’s the main outlier. Nikon uses mostly Sony sensors in their high end stuff (that’s D7000, D7100, D800/810, etc.). My only point is not to ignore potential differences as saying that saying “everything else being equal” when discussing sensor size affecting image quality is really all not that far fetched to mention. It’s only only correct from a theoretical point of view but pretty ball park of the market right now.
Personally, I find a simple trip to DXo and not reading their global score but the individual scores gives a pretty quick idea where the sensors fall relative to each other, as I have corroborated by different cameras, sensors and brands with their findings.
Basically saying as a rule of thumb that a bigger sensor will provide or will probably provide better IQ isn’t too wrong. And while you mentioned in a couple of replies “I never said that smaller sensors did better” the statement “Explaining why sensor of one size has a cleaner output when compared to a smaller sensor just because it is physically larger has one major flaw – it is actually not true once you factor in a couple of variables: sensor technology, image processing pipeline and sensor generation.” to me doesn’t seem all that valid in light that if those discussing say “everything else being equal” considering very much so that right now a lot of the market has a lot of equal in the comparison.
Now, if we were talking about say, end result by a JPEG engine, I would agree there’s too many variables here indeed – and that can be seen by the varied results even using the same sensor (Pentax vs Nikon is a good example).
Perhaps I should paraphrase that particular sentence, so that it does not sound as conflicting. I brought up “everything else being equal” later, so that particular sentence refers to a situation where one might think that a larger sensor will always be better, irrelevant of sensor technology.
Yes, re-read the article and correct all the flaws, or simply delete it.
Raist3d, so that’s where you and I are looking at things from a different perspective – you say “it is close enough, so it is OK”, whereas I say “close enough is not good enough to be valid, because things could change”. If that’s all there is to our disagreement, then we can just move on, as there is no right or wrong here, depending on how you look at it 🙂
To try to be more clear again:
“Raist3d, but wouldn’t you agree that basing a theory that is now being applied mathematically by some people to derive equivalence on a sensor technology that could potentially change any time in the future is flawed?”
No, if you are explicitly saying “everything else being equal” because that’s after all, theory- if you are talking about making a theory.
Agreed, if assumption is always made on everything else being equal.
And let me point out some of the issues I have with the article- for example in point 7: “So let’s get another fact straight: smaller format lenses have exactly the same light gathering capabilities as larger format lenses at the same f-stop, for their native sensor sizes. Yes, larger aperture diameter lenses do transmit more light, but more light is needed for the larger sensor, because the volume and the spread of light also must be large enough to cover the bigger sensor area. The Panasonic 42.5mm f/1.2 may behave similarly to an 85mm f/2.5 lens in terms of aperture diameter / total light transmission, field of view and depth of field, but the intensity of light that reaches the Micro Four Thirds sensor at f/1.2 is very different than it is for an f/2.5 lens on a full-frame camera – the image from the latter will be underexposed by two full stops.”
There are several issues I see with this here. One- since we are talking in the context about how larger sensor and gathering lights aspects- you are leaving out that while the bigger sensor here as you say has to spread out the light from the lens gathering more light, it ends up being a higher quality sampling of the light. Either the photo sites are bigger at the same megapixel count of the smaller sensor or you have more sampling points (at a higher MP count). If the photo sites are bigger you have more capture area of photons per photo site vs/ circuitry overhead than the smaller sensor.
Two- it is true that the Panasonic 42.5mm F1.2 lens behaves similarly to a FF F2.5 85mm lens and that the FF will be underexposed by two stops- but it doesn’t mention that the FF in this case can afford the difference in the under exposure by the higher ISO quality *and* the FF can give you the option of an F1.2 lens that would require 2 stops faster to match on the Panasonic (which doesn’t exist).
Both set of statements are valid- but one seems to come across as if somehow on the Panasonic end things were just equal and that’s it. That’s not the only thing that is said in these conversations. I do agree with the statement you made on the lens though and I do agree it’s good to keep in mind, but it’s not the complete truth imho (not saying you were trying to purposely mislead).
Now we are talking, thank you!
In regards to your first point, I agree and I believe I actually discussed that particular scenario further down in the article (in simplified terms) in regards to pixel size and resolution. Perhaps I should go back and address this in the same paragraph to avoid confusion.
In regards to your second point, I definitely addressed that as well when I talked about the fact that people talk about f/0.7 lenses that do not exist, and how hard it is to make such a huge lens for any format. I did not specifically talk about ISO on FF being better though, so I will definitely add that – thank you for reminding me of that (will have to figure out where to add that).
Here the non existing F0.7 lens, you can actually rent it, ask google.
https://celluloidpopculturejunkie.wordpress.com/tag/zeiss-50mm-f0-7-lens/
Educate yourself before you come up with claims about things that don’t exist!
Hopefully this is easy enough to understand and adds enough to what I said “It is important to note that although the comparison above is valid technically, a larger sensor would yield cleaner images and would allow for faster and less expensive lenses, as pointed out earlier.”. Let me know what you think – would be happy to revise if you come up with better and not overly technical language 🙂
Larger sensor is further from focal point, hence receive less dense light if aperture would be identical size with smaller sensor, that is closer the focal point.
Example: Make a pinhole camera and it becomes very obvious. And then you understand why you need larger aperture (pinhole) to maintain same exposure time and angle of view as with smaller “sensor” that is closer the hole. ISO can’t be changed as it is the print paper and always same. Exposure time is managed by covering the pinhole. And aperture is controlled by making the pinhole larger or smaller (larger means circle of confusion will be larger, leading to blurrier photo).
You will quickly learn why we have lenses, instead just one small hole. As we want a light density to be even between the whole paper area. Otherwise we get heavy vignetting, as the center of the paper gets denser light than areas around the center point, based to inverted square root ratio from pinhole to center of paper and paper edges.
If we have two identical size paper sizes, and want wide angle pinhole camera, we need to get the paper as close the pinhole as possible. The exposure times are very short, vignetting is very heavy but we have very wide angle image.
If we want telephoto pinhole camera, then we need very deep box where we can place the paper very far from the pinhole. The exposure times are very long, vignetting is minimal but we have very narrow angle of view captured.
That with two same size papers with same light sensitivity.
You could make the paper as large as you want for the “telephoto pinhole camera” but you would not get any shorter exposure time. Instead you would gain wider angle of view and raise amount of vignetting.
If we would get a half-sphere paper back of the pinhole camera, we would eliminate the vignetting as long the each part of the paper is as far away from the pinhole. But we would have a problem like what mappers do have, how to project a sphere on flat paper?
We could bring that “half-sphere paper” closer the pinhole to get faster exposure times, but same time we would need to keep the ratio of sphere to pinhole identical, meaning we would need to make a much higher angles for the paper and more twisted presentation of the view and much smaller piece of paper.
It is very basic geometrical problem that can’t be overcome with current known technology.
Even today we have round hole, lenses to distribute the light evenly (balance its density) for square flat area behind it.
Larger the area, larger the camera (camera is SLR body + objective, not just the body as that is like photopaper, camera is like pinhole camera where you have hole (objective) and light sensitive paper (film/sensor) and that in a box.
I think that the point being made is good, but the argument being used is flawed. As somebody else has already commented, compare a full-frame sensor and smaller sensors made with the same era’s technology and the FF sensor will usually be superior.
I’m also really surprised that no mention was made of the size of the photo sites. Compare a 16M pixel M43 sensor with a 16M pixel FF sensor and each photo site on the FF sensor will have 4 times the area of the photo sites on the M43 sensor. Larger sites = less breakthrough between sites = less noise.
Fraser, I think the fact that the site admin took only one section of the article is what is creating all the misunderstanding about what I wrote. I never said that smaller is better – there is not a single reference in the article to that. I stated that total light equivalence is flawed and gave my reasons for it…
Very true. Context is everything. 🙂 (Note to self, head back and read the whole article).
Fraser, I would love to get your feedback, but only once you read the whole article. If you find anything you disagree with, please bring it up and I would love to discuss in more detail…
I read yeasterday the whole article (in one stop) and there were few mentions I would like to say.
From this point as I don’t have now much time and starting from start, I would point out that:
– The reason to make a APS-C size sensor was not about price, it was about technology being so inferior that it was not possible to make larger sensor where electrons could be read out with long distance and manufacturing process to get a sensor that had no faulty pixels. The technology wasn’t there to make a larger sensors. The digital sensor technology started from small sensors first, and when technology evolved, we got bigger and bigger sensors. First there were problems to make small enough pixels, get the pixel pitch small enough etc.
– The APS-C is very confusing for many as it doesn’t really mean anything. But some photographers do remember that on the film era when the 35mm develoment was cheaper than 120 roll per frame etc, there was a new line up cameras using same 135 film but using smaller exposure box. Olympus PEN was such camera. It used half the frame size, hence the name “Half-Frame” and “Full-Frame”.
Photographers could use the same 135 film and then use different camera, the film came in rolls lengths for 24 or 36 frames (unless you rolled film yourself in darkroom and used bigger roll cassettes) and by using a Half-Frame camera gave possibility to expose 48 or 72 frames. The quality was more than enough for common 10×15 prints but the development prices were identical to 24 or 36 rolls as the photoshops didn’t bill about frames but about the roll.
The chemicals used in photoshops were required to be swapped after X amount of film was processed, it was counted in meters. So shorter film was cheaper than longer. Same way the narrower film was cheaper than wider (135 vs 120).
When the digital DSLR era came up, the sensors were Half-Frame size and that is APS-C format. But when the Canon and Nikon made their first 35mm sensor, as many photographers rejected the idea to not use their “native objective focal length” then Canon and Nikon marketed their cameras as “Full-Frame” as many knew that APS-C was “Half-Frame”. But because digital era didn’t require counting frames on roll, swapping film for required ISO, the “Half-Frame” term vanished pretty quickly. The “Full-Frame” term standalone as professionals were promoting it as Canon/Nikon wanted.
The difference just is, almost every digital sensor is “Full Frame” (expect those few multi-ratio sensors thats corners are always cut off) as it covers the image circle. There isn’t same thing as film, where we could use same film between two cameras exposing “Full-Frame” or “Half-Frame” sizes.
For long time Medium Format was the standard, and it didn’t change until news reporters required to have a simpler camera gear and papers wanted to lower costs developing film, 35mm format got its change. The 35mm film was the “Small Format”, and compared to Medium Format it really was such. For many photographers the Medium Format or even Large Format was the only way and not puny Small Format.
The photographers promoted that you needed to spend time to find composition, time the exposure etc and Large and Medium Format allowed those with Technical Camera rigs. The Small Format was like Smartphone Cameras are today. Demoed for “house wives” and “business mans”. Even the smaller format tried to push trough but didn’t succeed because development was expensive by labor demands.
When much later Small Format hit USA (it SLR was long time used in Europe before USA) it really became a “golden standard” as now “everyone could” make prints at home, get 24 or 36 frames that covered almost whole family holiday! Safe and easy rolls etc.
Everyone who talks about “Full Frame” is talking about digital cameras but with film cameras terminology.
In one way it fits, as there is a trend in some digital cameras to allow user to use “APS-C compatibly mode” like Sony A7 series or Nikon etc. So photographer can turn “Half-Frame” mode instead using full frame, meaning they use half of the sensor. Just so they don’t need to do crop in post process when using APS-C objectives.
Does A7 lose a stop of light when photographer hits “APS-C compatibly” mode? NO! No “Full-Frame” fan who claims smaller sensors to receive less light have managed to proof how suddenly the A7r sensor receives one stop less light when they enable the mode, as the sensor disables electronically the not-used pixels.
– If the “Full-Frame” fans would be correct, they would need to correct the exposure after enabling the feature. But they don’t.
Same time m4/3 users would gain 2 stops more light when mounting a objective that is designed for Small Format, or gain even more if using objective designed for Medium Format! But that doesn’t happen. Metabone Speedbooster is required, as it recollects the light and focuses it to smaller area = densier light (like laser is about collecting light and focusing all of it in small beam to smaller area). Some objectives have it build in, like the famous NASA f0.7 objectives.
If a “Full-Frame” fan would build a pinhole camera and start doing some tests with it, it would be quickly found out that your article is correct. Sensor size only affects to angle of view, not to exposure. But there are other small things that are easily found out, why actually larger sensor require larger aperture and that means longer focal lengths for same angle of view and finally why they would otherwise receive less light if not using larger objectives.
If I would have time I would have written shorter reply and gone trough more than just couple first arguments.
You wrote well the article and it was nice to read. Just hoping you would revise the last chapters and little spend time polishing it.
why is everyone here so afraid of FF sensors? I really wanna see Olympus coming up with a FF camera next year and then see what happens to the die hard m43 fans, I think several of them are simply going to suicide or just break their necks and weak hands when they have to carry around 100 grams more…
They aren’t “afraid” of FF sensors, but rather do not like the size, weight, and cost of associated equipment that comes with a FF sensor. It isn’t just a few grams, but rather kilograms with a full kit. You can get a Panasonic GH4 with a 14-600mm equivalent set of zoom lenses for less than $5K, all capable of f/4 or less and they will all fit into a moderately sized camera bag. Try that with FF. Not gonna happen.
Also, as noted in some of SpeedBooster’s promotional materials, “squeezing” the image down to a smaller sensor (more accurately, smaller pixel sites) increases sharpness (as it does when you downsize an image on screen).
On the other hand, there is that pesky 2 stop difference in noise performance when you go from FF to M43… 🙂
Nasim- I understand that but it’s still a point you made and in that very point I see two things that are different being compared. I mean the statements are there as written. Where I do whole heartedly agree with you is that people who discard smaller sensors automatically- and by that deriding them, putting down- even when it’s not their solution, worry too much or don’t understand a certain key aspect of the photographic endeavor (and by smaller sensor I mean all the way to say the Q’s, not just CX or m/43rds).
You underlined well the points (and you are correct, that it doesn’t matter and I don’t know how people gets that you said X is better than Y) but I got the feeling as you wrote about posting it, that you got little tired at the end and “rushed trough” (still correctly) some last points.
As you wrote that you have not used film camera (do I remember correctly?), I would say that there is a reason why photographers are recommended to start with a pinhole camera. As that forces them to learn by the hard way the difference between larger format and smaller format.
Today very few have change to use technical camera, the most basic camera but most complex to learn and master. And again with technical camera we can use any size format we want.
While people don’t anymore do physically the prints at their closet or cellar, they don’t really grab “the process” from start to finish. Instead it is rubbed at their face about megapixels, sensor sizes, zooms, memory cards, touch screens, social networks etc etc.
I agree completely. Noise is inversely proportional to the sensor size. That is why 36mp D810 FF sensor in theory should have almost 2-stop better noise ratio than a 16mp m43 GH4 sensor, but in reality, we could never see the 2-stop difference in actual print (depending how large a print) especially in low ISO.
You’d only see a fraction of the noise in print, regardless of sensor size because print has a very limited dynamic range compared to a screen. Print would not be the appropriate medium to do critical viewing of a camera’s abilities.
Full screen viewing on a 20″or larger screen would show the differences if the lighting conditions were sufficiently difficult.
I agree that at relative dim lighting, the difference is approaching the 2-stop limit, but on most occasion it is very much less than 2-stop difference even it is view on a 20″ calibrated screen. I see more difference on the resolution and dynamic range.
I’m not sure I understand why you say that you wouldn’t be able to see a 2 stop difference. Go to DPReview and compare the two cameras you mention and I’ll bet you can clearly see the two stop difference. If it’s noticeably less, it is probably more an issue of pixel sizes than purely sensor size. An issue I mention in an earlier comment.
Yes, I mean the pixel sizes instead of the sensor sizes. Thanks. I really do not see the 2-stop difference from ISO 100-400 on my cameras and DPReview pics.
Yeah, that is probably because at that low an ISO, the signal to noise ratio is so high that it is not visible to the naked eye. If there is sufficient light, even very high ISO pics can look good because the signal to noise radio is so high. In another comment I think I mentioned this.
“Print would not be the appropriate medium …”
Let me turn this around: I print my pictures. So: I don’t have to care.
Fine with that.
Yep! That is why even compact cameras can produce great printed pics. The difference between DSLR and lower end cameras on a printed pic is color fidelity and often the 6 inches behind the camera.
Print IS THE RULING FACTOR for everything!
Circle of Confusion is ruled by the Print Size!
Depth of Field is by so ruled by the Print Size!
Sharpness is by so ruled by the Print Size!
And do you know what is the ruling print size? ANYTHING YOU WANT!
It depends from the print size, viewing distance, and viewers eye sight.
If I make a ruling factor for the “print size” my 4k display, I know the display physical size (28″ = 40″ x 22.5″ panel) and resolution (3840×2160) and pixel pitch (0.160mm) and so on pixel density (157PPI).
Now if I look my display from normal range that is 75cm, I observe photo to have a deeper depth of field and be sharper, than if I would look it from 30cm distance, that would make photo look to have less depth of field and less sharp.
If I backoff to 2 meter distance, the photo is even more sharper and has deeper depth of field than from 75cm distance (that is 0.75 meters for those who doesn’t get it).
Now if I send the photo to 55″ 4k display, that is 1.96 times larger, or as said 4 times larger by surface area. Then I need to look the display from two times from the distance to get same effect as with 28″ display.
150cm instead 75cm
60cm instead 30cm
4 meters instead 2 meters.
So while the “print” is LARGER, I can sit further from it and see it as sharp as sitting front of my computer.
Now if I would have a bad eye sight (near sighted) and I couldn’t see well further than 1 meter, that would mean I see the same photo in 28″ display from normal viewing distance (75cm) as sharp and shallow depth of field. But if I would look it from 55% display from same ratio range (that is the 1.5m) then I would not see it as sharp and I would see it to have very deep depth of field as I don’t spot the difference in sharpness.
– SAME PHOTO (same file)
– TWO DIFFERENT physical SIZE (same resolution, different PPI)
– viewed from TWO DIFFERENT DISTANCES
= Results to same or different experience based what is the viewer eye sight.
Now if I would use 2.1Mpix m4/3 camera and then 8Mpix Small Format camera (that is 35mm, if someone didn’t know) and I would use objective that gives me same angle of view (like 50mm and 25mm) using same ratio capture.
I could show the m4/3 photo in 28″ screen that is FullHD capable (1920×1080) and Small Format photo in 55% screen that is UHD capable (3840×2160) and then I could watch both photos from same distance and see that they are as sharp and with same depth of field (identical) but just physically different sizes (two times).
Now, to hang the 28″ photo on my wall takes some space. But the 55″ size photo would take a lot of space.
With 28″ frames I could fill a wall with half a dozen photos, but I could hang only 1-2 55″ photos.
If the wall is on stairway, there isn’t much space and people walk close by, so the photo needs to be small. Such condition the 28″ is better choice than 55″ size photo.
But if the wall is on living room where there is 5m distance to wall and it is hanged over dinner table, then 55″ framing size is probably a better choice.
So when a customer calls and order a specific size photo (largest I have done this year are 3×5 meters) with specific theme. I need to know where it is going to be hanged so I know what I need. The print size and viewing distance rules what aperture I need to use, so I get enough depth of field to get the subjects in the framing enough sharp.
It isn’t enough that I look the photo from camera rear display to confirm is it sharp or not. I need to do the math. And for that I need to know what is resolution of the sensor (megapixels) and sensor size (how large enlargement I need to do from original to final print size) and viewing distance.
You don’t roll camera aperture and say “okay, f8 is enough for this” and take a shot, as the viewer experience what is depth of field and sharpness is totally dependent from the final “print” size.
When we do large prints, m4/3 and Small Format (35mm) doesn’t have noticeable difference until we actually hit to situation where Medium Format is required as Small Format doesn’t anymore cut enough difference. There is a difference between m4/3 and Medium Format, but not such difference between m4/3 and Small Format. There is a difference between Small Format and Medium Format, because we are talking about very large prints that are viewed from very close range.
If someone does prints at home (A4/A3 sizes) then almost ANY camera is basically enough for them. If you just send photos via email, upload to social networks or just print photos to photo album, there isn’t any kind difference between formats as viewed size, distance and display resolution more likely is so small, short and low that it doesn’t matter.
I was discussing noise characteristics and whether print was the appropriate medium for a critical assessment of sensor noise.
Just to be clear, by print I’m referring to printing it out on a printer… only. I am also referring to the difference between subtractive (reflective/printer) and additive (projected/screen) light/image protrayal.
Because of the nature of print, it is almost impossible to reproduce the dynamic range you see on a screen. In addition, during the conversion from RGB to CYMK, detail is lost. For these reasons, almost all of the luma noise and most of the chroma noise is reduced in the conversion.
On the other hand, on a screen both the luma and chroma noise are retained and very visible. Also, since the VAST MAJORITY of pictures (at least 95%) are viewed on a screen, it is the most appropriate way to critically assess sensor noise.
Thus, a screen, not a PRINT (i.e. output from a printer) is the appropriate medium to evaluate sensor noise.
In terms of practical noise (ie: what you would actually see once you normalize the images), a FF sensor with more pixels but smaller photo sites can have the same overall noise characteristics as a FF sensor with less pixels but larger photo sites. Just look at the A7R vs. the A7S. The A7S doesn’t actually noticably pull away from the A7R in terms of ISO performance until after ISO6400. This is because the A7R has 3 times the pixels to work with and thus, when you shrink it down to 12MP you are also shrinking down the noise grain.
The author of this article did actually address this stating that, in many cases, having more pixels on the same size sensor is more like a win win, because at 100% crops at low ISO the higher MP sensor will have similar noise levels but will have more pixels to resolve more detail (or for cropping). And when the ISO begins to climb, you can resize the larger MP sensor so its noise characteristics keep up with a lower MP sensor. This obviously only works so far, and at some point on the ISO scale the larger photosites will make more of a difference than the larger MP count. But, as I stated above, getting equal noise performance up to ISO6400 (in the case of the A7R vs. A7S) seems worth it to have more detail/pixels to work with.
The problem in this article is that people who have extremely little knowledge of physics dare to discuss and put up theories on things they barely understand.
Man, before claiming someones incompetence, how about dealing with your own? At least try and properly identify the science: optical formulas are pure math, not physics. I guess another point goes to the US of A education system.
Optics is a subset of physics, Einstein. 😉
Unless the laws of physics enter a camera forum where they get changed, juggled and canonized to fit a believe-system… 🙂
Optics – yes. But not the optical formula. As long as you do not actually try to design a lens, it is pure math.
To me, a person with minor in math, this is probably the most ridiculous aspect of the equivalency discussion: literally nobody has a clue what they are talking about. Those who get a clue, soon after stop commenting on the topic.
Wikipedia has all this information lied out, with relatively simply formulas and accompanying geometric presentation to help understand the formulas. Yet, instead of spending 1-2 hours of reading through them and grasping the formulas, the relations, the numbers, the ratios, literally everybody spends days, weeks writing emotional posts about it.
That’s also why you would never get an opinion of or a comment from a pro scientists. Because there is nothing to comment on, there is no place for an opinion. For a person with math or physics education, it’s trivial: it’s only 2-3 stupid formulas.
Doesn’t it just take the speed of light to fix the total light issue?
Saying a smaller sensor will sense less light sounds to me like there’s
going to be too much light if you throw all that condensed light on it,
that would otherwise have equivalently been consumed by a FF sensor.
The
other way around is also a bit like explaining about the flat earth,
saying it takes the sun much more time and effort to light up the area
that represents a FF sensor than it would take to light up a smaller
sensor.
The only thing you can say is that there is usually more
space between a FF lens and sensor than there is between a smaller
mirrorless lens and sensor. Yes, the flange distance is another variable
factor.
Anyway, I’m not going to build a sensor and put it in a camera, I’m just going to take pictures.
In easy terms… which has the better vision: the elephant with it’s large eye (lens) and large sensor (brain) or the eagle with it’s small eye (lens) and smaller sensor (brain)?
All you confirm is your own miniscule brain size.
Sounds like BS to me.. written by 43 fanboy.
Let’ go take sum pictures! 🙂
We have a whinner !
If m43 had more modern sensors than FF or APS-C, this point would be relevant. But generally, there is very little difference between the most recent sensors for all three sensor sizes. Throwing in MF is just a diversion tactic, the sales volume of MF is so small (tens of thousands) compared to APS-C (millions) that deployment of technology is noticeably affected. And m43 probably only sells a tenth or less of what APS-C sells, making it closer to FF than to APS-C in sales numbers.
“sales volume of MF is so small (tens of thousands) compared to APS-C (millions)”
Source: El Aura’s imagination.
MF = medium format
m43 = micro four thirds
I was even to generous with my recollection of MF sales, the MF is estimated to sell 6000 units in total per year: http://www.forbes.com/sites/marcbabej/2013/05/08/how-leica-camera-is-reinventing-the-medium-format-market-on-its-own-terms/2/
To the non-believers…
If you get a 50mm ƒ/1.8 lens (film) and put it in a 4/3 camera, do the aperture blades suddenly become narrower and smaller?
Wow. News to me!
But you collect only 25% of the information with a 4/3 sensor. The information to create an image is contained in the photons reaching the sensor. If your sensor area is only 25% of that of a 35 mm sensor/film, then only 25% of the information can be recorded (the rest is discarded by absorption on the black interior surfaces of the camera.
Photons ‘paint’ an image like one could ‘draw’ an image with a paintball gun. If you draw an image with only a few dozen paintballs, you’ll get a dithered image. The more paintballs you use, the more precise and less dithered an image you can create.
The main argument of the article is logically completely flawed.
I will stop liking/subscribing to 43 Rumors for delivering (and advertising!) such low quality content and wasting our time.